i ľ The Differentiation of Society Niklas Luhmann TRANSLATED BY STEPHEN HOLMES AND CHARLES LARMORE Columbia University Press New York 1982 7 Politics as a Social System At the present time, there is no uncontested theoretical foundation for research in the area of governmental politics and administration. The time is past when practical philosophy, which took ethics and natural law as self-evident points of departure, shaped what questions were posed and what answers were considered. "Power" is now used only hesitantly as a fundamental concept and then only for a partial aspect. "The state" has remained a very vague category of little analytical use; it runs the danger of being amplified and constricted by various traditions and prejudices. "Government" is subject to the opposite weakness; by being viewed more precisely as an institution or organization, it no longer is understandable in its own terms. It seems that for some years the concept of a political system has been forcing its way into this theoretical vacuum and spreading rapidly for want of competing concepts. This is happening chiefly in the form of model building, developing different kinds of schemes for describing the structure and course of political processes (Easton 1965a; Almond 1965; Wiseman 1966; NettI 1966; Senghaas 1968). Reflection upon what way political systems are indeed systems has only just begun (Easton 1965). It has not yet adequately clarified the advantages and limits of this conceptual approach. And it has by no means made use of what it could learn i POLITICS AS A SOCIAL SYSTEM 139 from general systems theory, cybernetics, the sociology of social systems, the sociology of organizations, ethnology, psychological systems theory, and many more. At this stage of research it is not yet possible to pass a well-founded judgment on what the concepts of systems theory will be able to accomplish in this area. Opinions one way or the other will have to be preliminary, if not premature. It is thus all the more attractive to apply some general ideas in systems-theory from various disciplines to political systems. In this way, raising the level of abstraction will allow the interconnections and consistency of various premises, concepts, and forms of research to show forth more clearly. In this essay I shall attempt to show that (1) the differentiation of the political system as a particular realm of action distinct from the rest of society and (2) the great autonomy of this system within society allows it to develop (3) afunctional specification in issuing binding decisions and creating social power. Together with this differentiation there is an increase in (4) the internal complexity and selectivity of the political system, and hence (5) an enhancement of its capacities for communication through power becoming generalized and reflexive. This makes (6) an internal functional differentiation of the political system necessary, as well as an ability to take over problems from elsewhere in the society. In particular, this requires (7) internal safeguards against risks. Taken all together, these different variables explain how (8) political functions can be stabilized at a higher level of social complexity. I. Differentiation We can speak of there being a social system only to the extent liiat the system can be distinguished from its environment. JVfore-3ver, the agents themselves, and not merely sociological theory, must be able to recognize the boundary between system and environment.1 To the extent that this is the case, the system is "differentiated" from its environment. Whatever happens in its environment does not automatically happen in the system as well. The differentiation of social systems from their social environment can arise at various levels of meaning. It can apply to actual POLITICS AS A SOCIAL SYSTEM 14 persons, assigning them completely to only one subsystem.2 It cai apply more abstractly (and this is the general case) to particula roles or even only to particular goals and norms or individual values The differentiation of the political system takes place primarily a the level of roles. The individuals who perform political and ad ministrative functions cannot have their whole lives separated fron the rest of society.3 Bole differentiation thus leads to a correspond ing differentiation between decision-programs and values. But thi latter form of differentiation is not by itself structurally sufficient since the political system applies legal norms and represents value, which always claim to be valid in and for society at large. Only a the level of roles can this process of differentiation be carried ou unambiguously, so that it is usually recognizable whether a roh (e.g., civil servant, member of parliament, party secretary, voter, o petitioner) can be attributed to the political system. Frequently, this kind of differentiation is characterized as "roíť separation." But that can lead to misunderstandings. In reality, i is a change in the way roles are combined to support socia structure. In simple societies the certainty of expectations rests on a per son's playing a combination of roles. These societies, too, exhibi a number of different kinds of roles, but they are so minor that fhť same people are always meeting each other in the different roles Hence, expectations, mediated by personal identities, are trans ferred from the context of one role to another. A person's behavio in one role provides grounds for inferring what his behavior ma; be in another. One who fails as a neighbor cannot be trusted as í warrior. One who takes a false step in opening a dance can endange a planned marriage. One who is not economically independen cannot gain any political influence. In this kind of social orde persons and roles cannot be completely separated. A person mus fit a social stereotype, and the interplay of roles has to remair dependent on persons. Social control then influences behavior pn mařily through each individual's always taking his other roles intc account (Nadel 1957, p. 63f.). In differentiated systems, on the othe hand, the execution of roles is regulated by the behavior amon$ role partners; it is made independent of whatever other roles í partner fulfills. Roles inside and outside the political system the> POLITICS AS A SOCIAL SYSTEM 141 become mobile in relation to each other (i.e., they come to vary independently of each other). Whether politicians or civil servants are rich or poor, into which families they marry, to which gods they pny, which circles they frequent, with whom they are friends or nc ighbors, with whom they have been prisoners of war, and to w horn they owe money—all this is not supposed to be of any struc- * ' il importance. At most, it has only a tactical value in establishing i lations." The political system has to be able to accept that such • lations" in personal backgrounds change with those who hold ii> roles. It has to be indifferent, in its ability to survive even if 1 ' in its tactics, to the interweaving of roles in society, while at *b same time treating and exploiting them as lucky opportunities. f'rily ki this way can society be made the object of politics and iO ninistration. I'his restructuring will at first be seen as a loss of aid in social ii1 mtation, leading necessarily to a corresponding loss of security. Mnivever, individual roles will still be able to be combined tem-•> i ally, materially, and socially. In order to ensure this, they must '» ariented toward universal criteria, independent of persons and * ľ ir particular relations. This is best achieved through a causal n'her than cosmic or symbolic) understanding of human action. 'í iction is understood as causing effects, a process of rationalizing - »I is is triggered. Such a process emphasizes how every action t ses specific effects, while also requiring abstract norms and ■lues, and ultimately pragmatically handled ideologies, to provide f riteria for deciding who has to take on which role for the sake of ^ 1 ich effects and at what time.4 This requires the morally sanc-••oiied attitudes of objectivity, justice (in the sense of the principle if equality), and impartiality from those engaged in politics and administration. With this, the requirements for recruitment into nnlitical and bureaucratic roles are also changed. Individuals no linger need or can rely on ascriptive criteria predetermined by hii*h, nor on roles outside the political system. Instead, they must dťtelop criteria of ability and efficiency which are appropriate to thf system (naturally "good relations" can prove important here!).5 Pii determined social status also loses its importance: saddlers and •winters, corporals, and even those who have not served in the riilitary can be entrusted with political leadership. The domain of POLITICS AS A SOCIAL SYSTEM 142 possible recruitment grows, and thus the opportunities for selection as well. At the same time the political system becomes so complex that this alone offers protection against the sudden invasion of new forces. Newcomers have to work for some time at their careers; they have to be trained and tested before they can act Periods of apprenticeship are always periods of socialization as well: the requisite feelings and attitudes are imparted along with information and skills.6 Hence, mobility, selectivity, and rules of recruitment are indispensable in the far-reaching differentiation of the political system from the rest of society. II. Autonomy When differentiation succeeds at the level of roles, it paves the way for the relative autonomy of the political system at the level of decision-making criteria. The roles of the political system then interact within certain limits among themselves, developing their own regulative standards. In contrast to a mere differentiation of particular roles or particular criteria, the concept of autonomy has to do with self-determination—the ability to settle upon selective criteria for transactions with the environment and to change them if need be. Autonomy should not be confused with self-sufficiency. Self-sufficiency means isolation and independence from physical and informational aspects of the environment. Autonomy, by contrast, presupposes this kind of independence and refers to the degree of freedom with which the selective criteria of the system can regulate the relations between system and environment. Securing relative autonomy for the political system has been, from a historical perspective, an extremely long process of development, full of setbacks, which was first able to attain a position of stability in the modern nation-states of Europe. This shows that very complicated preconditions had to be fulfilled by the structure of society. Following some general ideas from systems-theory, which cannot be further examined here, we can distinguish three dimensions along which the conditions for autonomy lie. Temporally, autonomy depends on the environment's giving the system iOLITICS AS A SOCIAL SYSTEM 143 time to insert its own ways of processing information. Materially, it is important that the system, remain linked to the environment at two different generalized levels of meaning. And socially the differentiation of the environment into distinct partner systems is essential. In order to be autonomous, a system must first of all "have time." Il cannot always be forced to react immediately to outside impulses,7 instead, it must have the time to apply its own processes of "■electing causes and effects. This requirement is normally repre- ented by an input-output model which basically inserts the sys-ti'm's own processes between temporally separated causes and ,ef-lects in the environment (Easton 1965a; Almond 1965). By thus separating its boundary times, the system can use intended effects joals) to select the appropriate causes or imposed causes to select Mie appropriate additional causes and desired effects. Furthermore, Mie system then gains a certain freedom to switch these opposite perspectives as the need arises, and thus to choose the criteria of selection itself.8 However, the limits imposed by the environment on this passage of time let time within the system become scarce, ind this scarcity of time burdens both the formation of consensus ince that costs time) and the actual choice of programs and de- isions (whatever is limited becomes urgent whatever its value) Luhmann 1968c). In order to save time and use it rationally, a 'stem sets up its own timetables that cannot be fully coordinated * ith timetables in the environment. This again creates difficulties 11 ĺ adaptation. In a system's material relations to its environment, its autonomy depends especially upon these contacts being stabilized at (at least) t vo different levels of generalization. A political system must be more or less generally accepted in its social environment. As a ■< rstem it must enjoy political credit, which does not rest upon its uaking specific promises about what its decisions will be. This t edit is not created by a continued bartering, or taken back after i ^ery failure. Below this level of wholesale acceptance and the ■ithorization to make decisions, we find the level of concrete in-ractions, which fulfill the function of the political system. General cceptance and the symbols representing it regulate and limit what ik possible at the more concrete level of daily transactions. Seen POLITICS AS A SOCIAL SYSTEM 144 from the other side, there are thresholds built between these levels that block the transfer of breakdowns and disappointments from the more concrete to the more abstract realm. Mechanisms for absorbing protest, for example jurisdiction, have this function. The separation of these two levels is reinforced by the differentiation between important and unimportant topics within the system. But this separation seems especially to rest on the fact that for the acceptance of the system there exist different kinds of functional equivalents than those for the solutions to problems that are sought in concrete interactions.9 In the social dimension the most important condition for autonomy is that the political system not see itself faced with one solid power in the environment (e.g. a church or a big stockholder), but instead depend on several relevant environments. The differentiation of the environment can consist in a number of identical environments—e.g. the families, clans, and gentes of simpler societies—or today in a number of equally famous professors providing advice or a number of organizations giving access to the capital market. But the environment can also be split up (and this is more important today) into a number of functionally differentiated subsystems. Thus, today, for example, the sources of religious legitimation, the sources of reliable information, the sources of economic support, the sources of personally motivated willingness to take on roles all diverge. Of the corresponding subsystems in society it is only demanded that sector by sector they meet specific needs of the political system. The political system can then negotiate with each individual system by relying on the backing of other domains in the environment and balancing the way these various contributions are combined. This requires a rather highly abstract structure. Furthermore, care must be taken that the internal differentiation of the political system (see section VII below) does not come to correspond to the external differentiation of the environment. If so, the parts of the system will identify with environmental factors, or cross-boundary alliances will be so likely that the political system will lose its function of integrating society and fall to pieces.10 When we combine these different presuppositions of high autonomy within the system, we already have a very complex picture that we can hardly visualize. To every relevant section of the en- POLITICS AS A SOCIAL SYSTEM 145 vironment we must imagine the political system to have relations at various levels of generalization, and these relations must have different time limits for regulating input and output. The improbability of such a social order is obvious, and explains why it appeared late in history and indeed only once.11 But relative autonomy together with adequate differentiation is indispensable if the political system is going to be specialized for a particular function which has no parallel elsewhere in society. III. Functional Specification As we have seen, the differentiation of a system from the rest of society makes it possible to delimit and specify functionally the context in which persons anticipate the roles others will play, and the context in which they motivate and check each other. In this way the technical advantages of a division of labor can be gained. The specific function of politics in society can be analyzed so that we can discern its own laws and rationale. Religious, economic, cultural, familial, educational, scientific, and military contexts confront the political system as environments that demand appropriate attention but without determining what political and administrative actions are performed. In other words, by this process of differentiation and the con- ! dinons favoring high autonomy within the system, the political svstem is put in the position of being able to make decisions. The specific function of politics is fulfilled at the level of concrete interaction because these decisions are made binding.12, A decision is binding whenever, and for whatever reasons, it succeeds in effectively restructuring the expectations of those affected and thus becomes the premise for their future behavior. It is a matter, then, \ of a£tualle&r,mng3iand not just formal validity.13 A political system's potential for assuming the specific function of making binding decisions must increase as society grows more complex and increasingly poses problems that can no longer be solved by reference to truths or common convictions, or by mutual sympathy or exchange, but instead only through decisions. The differentiation, autonomy, and functional specification of a political POLITICS AS A SOCIAL SYSTEM 146 system are thus consequences of the increasing complexity of society as well as prerequisites for a further increase in this complexity. With increasing complexity, it is increasingly likely that the advantages of efficiency lying in functional specification will be realized. Two aspects of our characterization of the specific function of politics as the issuing of binding decisions remain to be discussed— those dealing with which decisions are made and the motives that underlie their acceptance. A political system that has been differentiated from the rest of society can no longer bind itself in those areas.14 It must be able to vary the issues it decides upon, depending on which problems the society places on the political agenda. Only then can it orient its decisions toward the fluctuation of problems in society. And it can no longer rely on particular motives for obedience that may lie in individuals, groups, or situations; for example, it cannot rely on a combination of physical force and loyalty, ideological convictions, or the calculation of economic gains. Instead, it must be able to combine and equalize incompatible motives of the most different sorts, so that an almost unmotivated and matter-of-fact acceptance of decisions can arise. Only through structural abstraction, which in both these respects ensures openness while keeping the system stable, can the political system be designed for the function of issuing decisions that will be accepted as binding.15 Each of these forms of indeterminacy would be difficult to achieve and maintain. The problem becomes even more acute when they are projected onto each other. This means that the political system must guarantee the acceptance of still undetermined and undefined decisions—in other words, the legitimacy of legality (see section VII). Any system that tries to accomplish this will have to maintain a sociologically rather improbable order.16 The fundamental question of modern political philosophy—under what conditions is such an autonomous form of political domination legally permissible?—must be reformulated into this question: under what conditions can a differentiated, functionally specified political system be stabilized within its social environment? Our discussion of binding decisions as the function of the political system has so far, however, only focused on the level of concrete POLITICS AS A SOCIAL SYSTEM 147 interactions and their premises. At a much more abstract level the political system has the function of generatingj^wijü powjzr. As a (X)__ political system is differentiated from the rest of society the amount V' of a\ailable power in all of society also increases.17 By power I ŕ mean (and this point I shall return to in section V) a medium of communication that allows decisions to be transmitted (thus, it is not as in classical theory, a single cause having specific effects). Through the process of making a specific power base abstract, which has generally been characterized as ensuring peace through an effective monopoly on making legitimate decisions about the use of physical force (Weber 1956, p. 29; Elias 1939), the political system is in a position to differentiate power. The means of physical force a're kept in barracks and thus cannot compete with other sources of power. That is why a number of different kinds of power bases can be tapped, developed, and cultivated. Formation of power by means of freely negotiated labor contracts or memberships in organizations is one of the best-known examples. Another is the power that comes from occupying a key position in communication networks. In this way the power available in all of society, and not just the power of the political system, can increase significantly while being related to very different constellations of motives, as befits the requirements of a highly differentiated society IV. Internal Complexity and Selectivity The most important condition for stabilizing a differentiated and functionally specified system is that it maintain its own complexity ata level corresponding to that of its social environment.18 By complexity, I mean the totality of possible events—the complexity of the world is all the possible events in the world, the complexity of a system is all the possible events compatible with the structure of the system (Luhmann 1967b). Hence, a high internal complexity entails allowing alternatives, possibilities of variation, dissent, and conflicts in the system. For that to be possible, the structure of the system must be, to a certain degree, indeterminate, contradictory, and institutionalized in a flexible way. Against the natural tendency POLITICS AS A SOCIAL SYSTEM 148 toward simplification and the removal of all uncertainties, it must be kept artificially open and remain underspecified. Outside of cybernetics, it has been seldom recognized that the creation of a tolerable indeterminacy in social systems is an achievement and not a mishap. (Beer 1962, p. 60f.). Typical complaints about the conservatism of bureaucracies (Laski 1930, pp. 70-73), the dogmatic tendencies of ideologies,19 the lack of political sensitivity in mass political parties (Lohmar 1963), or the decay of a real parliamentary opposition (Ellwein 1966, p. 212), continue to show that even comparatively mobile and changing institutions which are fraught with conflicts are being measured against a degree of complexity that is extremely hard to attain. But how does a political system arrive at an adequate level of alternatives and possible ways of changing? How does it avoid structural commitments which are far too concrete, which can block its maneuverability, and which—in a changing environment—can lead to crisis and breakdown? In particular, how does it keep the products of historical events from congealing into structures and eliminating other possibilities? From a general viewpoint, a system's complexity cannot be greater than its ability to reduce complexity.20 "Other possibilities" are not accessible to a system unless they can be meaningfully experienced, included in decision-making processes, and selectively eliminated. The more drastic the means of reduction are, and the more simply and concretely the decision-making processes are structured, the fewer alternatives a system can have. If it has to fit its environment into a friend/foe scheme because of a lack in sufficiently differentiated and generalized premises for processing experience, then its own behavior has already been fixed in its broad outlines and can vary only tactically. If it lacks internal processes to regulate how it can vary its own causal contributions, the system can only "blame" forces in the environment for particular events. The result is a correspondingly simple image of the environment (Harvey, Hunt, Schroder, 1961; Schroder, DiVer, Streufert 1967). A categorical and moralizing form of processing experience becomes inevitable. In political systems this connection between complexity and selectivity is particularly important (Vickers 1959, p. 104f.). As sys- ■OLITICS AS A SOCIAL SYSTEM 149 ems, they are specialized in issuing binding decisions. They have to arrive at them. This is so much expected of them that even making no decision is imputed to them as a decision. Under these circum->tances, the temptation to use simple means is particularly strong. On the other hand, they, too, can achieve their own complexity, md keep it a feature of their structure, only if they succeed in -trengthening the selectivity of their processes. They must be able to coordinate decision-making processes so that the reduction of tomplexity achieved by one is not lost for the other and does not liave to be repeated. Instead, it will be conserved and can be used t n a meaningful way. I am not referring simply to the administrative technique of mak-'ng decisions according to programs. I also have in mind the po-I Lbcal system's own treatment of conflicts. A political system's ability í d absorb social conflicts has to increase when society becomes noie complex and conflict-ridden. The political system then changes these conflicts from being cases of outright opposition to l>eing cases of regulated, articulate struggles to influence the de- ision-making centers. The absorption of conflict depends on cases i f opposition being transported from the environment of the system *o its interior; they are then represented as internal contradictions *nd, because these new forms bring a different constellation of lotives to bear, they can be better solved.21 However, conflicts ..an be taken into the system and legitimated only if the complexity f contradictory demands can be channeled toward decisions. That, íoo, requires selective behavior to be coordinated. A real uncertainty about the eventual outcome must be structurally guaranteed .e., power has to be suspended) in order for participants to feel motivated. The participants must be brought into roles which require complementary action (e.g., asking and answering). And they will have to give binding accounts of their position—ones that their opponents can rely on. Only in this way can a controversy become the object of communication; issues can then be better defined and enough alternatives eliminated so that a decision made with minimal selectivity and having minimal social impact can result. A political system's potential for resolving conflicts must thus be *sen as the result of several variables. First, it depends on the ixterit to which oppositions and conflicts in society at large are POLITICS AS A SOCIAL SYSTEM 150 politicized, or thought to require binding decisions. Secondly, it depends on whether the political system has been sufficiently differentiated from the rest of society so that, in taking over a conflict as its own affair, it can activate new motives. Furthermore, the degree of abstraction and the scope that the structure of the political system has for defining the limits within which conflicts may be permissible are equally important. Finally, and most important of all, the system's potential for conflict depends on how selective processes can be coordinated with one another. V, Power Traditional political theory has used the concept of power to designate the point at which it has dealt with this problem of coordinating decisions. This makes understandable the good, if limited, justification there has been for moving the theory of power into the center of political science. The transmission of decisions is, indeed, decisive. Power is the possibility of having one's own decision select alternatives or reduce complexity for others. Nonetheless, the question remains whether the premises of the classical theory of power are indeed adequate, or whether—as James March surmised after long deliberation (March 1966)—they cast the analysis in a form that may fit small groups, but not highly complex systems. The classical theory is based not on the idea of selectivity but rather on a supposedly causal form of communication. That is, it tries to conceive of power as a cause that brings about specific effects, or generally as the ability to make others act in a way they would not have chosen of their own volition (Dahl 1957). This leads to a transitive, hierarchical conception of power, excluding the possibility of circular and reciprocal power relations. It sees power-holders' calculations essentially as predicting a causal sequence (hence, it presupposes the availability of complete information), that is oriented by fixed needs and by the ability to prevail in cases of conflict Furthermore, it understands power as a kind of possession that can be gained or lost and that remains constant within the Pol [CS AS A SOCIAL SYSTEM 151 ■>Wi in: every loss of power means a corresponding gain of power tor an opponent and vice versa. UI of these conceptual presuppositions have to be redefined if »t are to construct a concept of power on the basis of systems tlii ory It will be based on the selectivity of communication. Power tlit-n has to be seen as selection based on selection, or as the strengthened selectivity of the system. In comparison to other ways ■it transferring selections, its peculiarity lies in the selective action not just the experience) of the powerholder, who in turn selects :rii action of the subordinate (and not just the experience the subordinate will have). Power exists whenever a decision-maker hi loses one specific possibility from among many and when this *>* íecbon is in turn accepted by others as a premise for their own d* Lision making—even though it obviously is based on a selective Jti'ision, Because it is obvious that there are other possibilities, tin subordinates are in turn motivated to implement the decision. 1 heir acceptance of the decision takes place because they are aware "Iiiit permanently stabilized alternatives, unpleasant for all the par-*n ipants, do exist. One of the most important is the possible use ■1 physical force, but it is hardly the only one. Others are perhaps in» »re important today for the internal processes in politics and ad- ii i lustration: for example, the withdrawal of important partners in»m a cooperative effort, breakdowns in technically functioning *"■ terns, loss of membership in organizations, the discrediting of i u ious individuals, or an overload of complexity, uncertainty, and fponsibility. The possibilities for expanding and distributing ■ vex within the system depend on how the alternatives to be ^ ided can be combined or played off against one another. The 'i'-sire by all the participants, even if they have opposite goals, to nuintain the system remains a condition for making decisions.22 Tli« extent of power and thus the organizational demands it must mi st will vary with the complexity of the total system—i.e., with it1 number of possibilities. I f the complexity of the system exceeds a certain threshold, power must remain reflexive. That is, it must be applicable to itself 1 uhmann 1966b), even at the top of the system.23 As complexity im reases, powerholders can expand their influence only if they let POLITICS AS A SOCIAL SYSTEM \l 52 themselves be influenced and if this influence is consolidated into predictable structures. By themselves, they could not utilize the decision-making potential of the system or maintain the complexih of the system. In all fully developed political systems, the relaiive power of one position must be able to influence the relative power of others, if the ability of the decision-making process to make selections is going to be thoroughly strengthened. The flow of power in the system must take the form of passing on premises for making decisions. But that is possible only if the roles in the system have been differentiated from the system's environment—i.e., only if the role a person performs in this system has not been immobilized bj his consideration of the other roles he performs elsewhere. The roles in the system must react to influence according to rules that are internal to the system and govern long chains for transmitting decision-making premises.24 The reflexivity of broad power-processes underlies, in turn, a further aspect of the political system: its functional differentiation into subsystems that perform various subfunctions in issuing binding decisions.35 Only when the application, of power to power has been stretched across a large domain and yet remains guaranteed is it possible to build subsystems within the political system which while operating under different and incompatible conditions, are still able to issue decision-making premises for each other and can thus be integrated. VI. Internal Functional Differentiation We now return from our discussion of political processes to structural questions. But this time we are concerned not with the differentiation of the political system from the rest of society but instead with its internal division into subsystems.26 This discussion will round out our overview, allowing us to formulate the hypothesis that differentiation, autonomy, and functional specification revolutionize both the internal processes and the internal structures of the political system. They require not only an increase of selectivity, and hence power, in these processes, but also a functional differentiation of the internal structures of the various subsystems DLITICS AS A SOCIAL SYSTEM 153 Rather often the decisive criterion for progress has been thought i i lie in an increase of functional differentiation (Smelser 1959; aPalombara 1963, p. 39f.; Eisenstadt 1964; Parsons 1961a). And m fact this does seem to be a necessary condition for the civilizing svelopment of more complex social systems} a condition closely nked with all the other variables we have mentioned. But it is jually certain that this factor cannot be dealt with in isolation iliggs 1966a and b). Functional differentiation depends on envi- mmental preconditions that can be analyzed, and it gives rise to jecific internal problems that must be solved if a functionally ifferentiated structure is to be formed and maintained. Roughly speaking, in modern political systems a fundamental ifferentiation of roles for bureaucratic administration, party poli- cs, and the public seems to take shape—historically, more or less i this order.27 Bureaucratic administration (in the broad, functional ;nse that includes parliaments and courts) becomes specialized i elaborating and issuing binding decisions according to politically redetermined criteria; in addition, these decisions must satisfy irtain requirements of mutual consistency (i.e., legality and econ- my). To this end, members of the administration (in contrast to pol- 1 icians) are given ministries with specific portfolios. To the extent tat the bureaucracy's criteria and programs for decision-making ecome variable—and especially when law is made positive ^uhmann 1967a)—it becomes necessary to set apart a particular Dhere of politics, having its own organization and form of work, r» deal with forming political support for programs and decisions. or as decision-making premises are allowed to vary, the conditions ! ir political support begin to fluctuate. For the realm of politics in (•us narrower sense, specific criteria of rationality (e.g., rules for tinning elections) are formed to different degrees in different sys- Wms These criteria do not need to conform with the morals of all »if society or the specifically bureaucratic ethos of consistent de- ision-making. Politics thus becomes suspect. Politics and admin- ■ tration are separated not only in terms of roles, but also in terms if their differing subgoals, which can no longer be integrated by jme common purpose, but only by built-in auxiliary conditions for •ecision-making. To the extent that politics and administration are •jparated by their different functions, a number of roles among the POLITICS AS A SOCIAL SYSTEM 154 public emerge on their own as a third element of the political system. No longer does the citizen, with his wants and his patience, stand as the subject of an undifferentiated authority. Instead, he occupies a number of special roles: taxpayer, proponent of resolutions, complainant, voter, writer of letters to the editor, supporter of interest groups, etc. These roles are divided up in accord with the requirements of the political system, and especially with its channels of communication. They are roles for complementary behavior and, as such, belong within the political system and not in its social environment. Hence, the exercise of influence comes to depend on behavior's conforming to these roles. When along with that a significant differentiation of these roles from the rest of society is guaranteed, much greater freedom can be accorded both to individual behavior and to the way an individual may combine these roles. Their specific function of converting experience with the results of binding decisions into new motives for decision-making can then be fulfilled largely within the system. But this will depend on the political system's being able to create enough alternatives and maintain a complexity that corresponds to that of society. The importance of this internal functional differentiation can be seen from a comparison with older forms of society, which were based on the opposite principle of segmented differentiation. These societies built identical subunits in the form of families and tribes. Segmented systems are made stable by the vagueness or even irrelevance of definite boundaries with their environment.28 They can lose or gain segments without altering their structure; family relationships become vague; domination is held in check by the possibility of seceding and by loyalities crossing its boundaries (Schapera 1956, p. 22f., I53f., I75f.; Gluckman 1955; Fallers 1963). In contrast, the stability of functionally differentiated systems is based on clearly demarcated boundaries that make it possible to distinguish which actions belong to which one of the various systems. They separate functional contexts and guard against interference. Fixing internal boundaries—as can be seen in the case of the political system—helps to fix external boundaries, since it enables the contacts between systems and their environments to be functionally specified. Politics and administration have their respective publics. Corresponding to their own further differentiation 1 JLITICS AS A SOCIAL SYSTEM 155 i ito parties or departments we find complementary roles differ-' i,tiated in their publics. Therefore, politics and administration, at í ast ideally, focus on society across defined boundaries. And only • 'ie public can legitimately consider its other roles in those of its i -tions having to do with politics or administration.29 Internal differentiation thus serves to specify and stabilize nmdaries. The differentiation of administration, politics, and the i iíbhc from the rest of society must therefore be carried out dif-irently for each of these particular functions. The administration i ust be separated, by means of its bureaucratic organization, from fl e civil servants' other roles in society; this is the main thrust of Weber's ideal type of bureaucracy. The differentiation of politics I om the lest of society is based, at the lower levels of party orga-jzations, on the same principle. At the upper levels it is carried ■ it by means of abstract conditions for electoral success—even if !• ss completely, since here other social roles (e.g., the victorious sneral, the astronaut, the professor) that create publicity provide * >od starting places for a political campaign.30 The public is dif-1 rentiated from the rest of society especially by virtue of its having < 'cess only to specific issues and channels of communication. For .ample, it can, in the role of voter, only choose among simple Iternatives, and in making this decision each individual finds him-lf "overdetermined" by his other roles—i.e., set free.31 These fferent mechanisms for achieving differentiation from the rest of tciety are coordinated so that the administration is separated from her roles more than from politics, and from politics more than i am the public. This is so because as the inordinate complexity of 3cision-making grows, the capacities of these decisions to be ra-inalized decreases. The complexity in the system also varies with the extent of its i iternal differentiation. The distribution of complexity, both in ex-1 nt and in determinateness, is directed toward the subsystems' i notions and capacities for decision-making. That is why selective I ocesses of communication, which first reduce complexity and ' en transfer it, must follow a definite order that is coordinated with the functions and capacities of the subsystems. Communication that is supported by the official power structure ■ >rmally flows from the voter, across political processes, through POLITICS AS A SOCIAL SYSTEM 15 the administration, to the recipient of these decisions. It seems 1 legitimate a reflexive application of power to power. It orders th communication process of the system by further specifying an reducing the complexity of the total system. That is why less d( fined decision-making situations are preferred to more define ones, and why the more unstable parts of the system guide the moi stable ones.33 The power of such a complex system, however, ca only be actualized if each powerholder accepts decision premise set by others, and subordinates himself to the power of others. Th greater and more indeterminate the complexity of his decision-mal ing situation is, the greater his power. There thus arises a con munication process in the opposite direction: the eventual recipien of the voters' decisions energetically makes his wishes know* through interest groups, the bureaucrats work out the possible de cisions for the political realm, and the politicians try to persuaď the voters to give their support In such a system, power must thu be exercised reflexively and reciprocally. That requires an exactin*. organization of the communication process, which gives rise in aľ the subsystems to power which is original, but capable of bein_ influenced, and which separates decision-making situations so tha reciprocally exercised power does not collide with itself. In thi*. way the official exercise of power and the exercise of power in th« opposite direction are held apart. The first has to be exercise legally on the basis of competencies, while the second has to b exercised informally on the basis of overburdening officials with complexity. Each direction of power is supported by different mc tives for its acceptance.33 VII. Internal Safeguards Against Risks As the complexity of social systems increases, greater risks or viously have to be taken—risks that can no longer be warded oil at the level of elementary face-to-face behavior or by reference 11 socially unambiguous dangers. Instead, in a very indefinite form "security" becomes problematic (Kaufmann 1970). This can eve i be observed under rather simple conditions—for example, when increasing complexity in the economy requires that even outsider*- 'OLITICS AS A SOCIAL SYSTEM 157 he given credit without being restricted by the diffuse and inescapable sanctions of a local community; or when trust and loyal support must be given to political rulers even if one is not related to them. The risk increases dramatically when a political system that is becoming autonomous demands that it be given sufficient time to act and total authority on decision-making. Such an authority «.an make almost anything possible since there are no longer powerful external controls that can keep decisions within the bounds of the desirable. In contrast to the old problems of the capricious and despotic ruler, which could be solved through admonition, education, or assassination, the political philosophy and legal theory of early modern-times focussed on the completely new problem of the uncanny arbitrariness of decision-making. This was a problem which, by being provided for structurally, was unavoidable. Not by accident, safeguards against this problem were at first still sought in law, in the limits of legitimacy, and in the establishment of controls. The fragility of these protective measures was obvious once law had become fully positive. Furthermore, it is questionable whether the problem has been fully grasped when it is viewed as the danger that rulers may misuse power. The risk that political support may •luctuate in an unforeseeable way is another aspect of this fundamental problem. In both cases it is the highly structured complexity , f the political system that suggests the use of all-too-drastic forms if reduction. The indefinite abundance of possibilities suggests the use of simple means. The structural conditions of highly rational decision-making processes having variable premises are equally conditions that reinforce the insane use of power or unleash uncontrollable political mass movements.34 A theory of the political system can approach these phenomena with a new conceptual apparatus and, without denying them, can draw out some of their other aspects. In contrast to the viewpoint of action theory, this theory views the problem, not primarily as the danger that power will be exercised in a morally reprehensible way by too powerful a ruler, but as the need to create and maintain alternatives in the system, to stabilize great complexity. That is why for it the danger is not too much power, but too little power in the political system.35 POLITICS AS A SOCIAL SYSTEM 158 The political system of a highly differentiated society can no longer be understood as a means to an end and can no longer be regulated by rigid external guidance. In order to pursue its function it has become so differentiated from the rest of society, so autonomous and complex, that it can no longer base its stability on fixed foundations, practices, or values. It can become stable only by creating possibilities for change,36 In this way, variability has become a condition of stability. Hence, it must be given structural guarantees. In other words, all social systems have the natural tendency to eliminate great complexity and insecurity by means of emotionally based identifications and fixations, personalized mechanisms of domination (based on personal connections and loyalty), tactical simplifications, and the construction of a historical tradition of generalized experiences. The structure of the political system must counteract this tendency.37 Incentives for continually creating new alternatives must be built into the system in such a way that they survive individual goal-oriented actions and can regenerate themselves. The system has to have command over possibilities for change, without thereby eliminating these possibilities by viewing what is new as definitive. Paradoxically, beyond a certain level of complexity, security can be achieved only by the detour of maintaining insecurity. Some of the typical institutions in modern political systems focus on just this paradoxical problem. It seems that, in order to solve it, they make use of procedures that presuppose the reflexivity of power, transmitting it to decision-making processes. The positivity of law, an orientation toward interests instead of truth, the pursuit of security through planning instead of tradition are some examples of this (Luhmann 1966a and chapter 5). Law in a society becomes positive when mere legality comes to be recognized as legitimate—i.e., when law is heeded precisely because it has been made according to definite rules by competent decisions. In this way, arbitrariness about a central question of social life becomes institutionalized. This arbitrariness is bearable only if there are rules guiding how it is made concrete. Furthermore, as a continuing structure law must become so complex that it cannot be totally or capriciously changed; it can be changed only by alterations in the given structure—i.e., only on the basis of the I OLITICS AS A SOCIAL SYSTEM 159 status quo. Both of these features of law together lead to an institutionalization of procedures. Procedures are a particular kind of t hort-term subsystem which reduces complexity and legitimates decisions. The distinctive feature of this institution lies in the fact that it continues to create open and conflict-laden decision-making i'tuations only as a temporary source of uncertainty; by means of tins uncertainty it motivates both onlookers and participants to take ■3art in the absorption of this uncertainty. Once sources of legitimation lying outside the political system can no longer be Presupposed, legitimation must be achieved within the system Jiself;38 Parallel to this normative determination of norms we find a similar degree of arbitrariness institutionalized in the evaluation of v alues. Underlying this development is the fact that values as well .15 the pragmatic reduction of horizons of action have lost their ' ipacity for being true,39 Now values must themselves be evaluated .tad revaluated according to their function in orienting action.40 In tlie political system truths are no longer at issue (although, naturally, firm everyday truths, such as that planes can fly or men must '«leep, remain in use). Instead, it is interests that are at, issue. Whoever still tries to advocate truths is making a mistake and is Sweated as doing so. The evaluation of values can be institutionalized in various ways—for example, in one-party systems with an undogmatic ideology which, for a privileged circle of participants, serves as a basis ľir discussion and consensus; or in multiparty systems with a sys-''m of political competition that allows a subsystem to treat values iiistrumentally in terms of their results (especially electoral results). In any case, opportunism becomes essential for survival. Values - in no longer by systematized by means of rigidly ordered priori-'ies. They have to continually win validity and urgency from one Mtuation to another, by being compared with how much other val-"es have been realized. A sufficiently broad basis for this development depends on a sufficiently large number of values in the wstem finding advocates and on relative preferences changing sufficiently quickly, In this regard, too, such a high degree of complexity can be mastered only if there can be presupposed a status quo which may be improved but must not be made worse.41 Fixing POLITICS AS A SOCIAL SYSTEM Ifi on a status quo appears as a new way of transmitting history, on which is favorable to planning. The shared assumption of consen sus speaks on behalf of the status quo. To that extent, the statu quo works as an institution: whoever wants change must assurn the burden and risk of taking the initiative. In addition, the status-quo can take the form of subjective rights, which can only be taken away in exchange for compensation. Or it can take the form o I performance quotas that should be surpassed or at least adhere< to. In both cases the past is used as a guideline for drafting thi future, but not as a legitimizing symbol. History no longer pose obligations as a binding tradition that shows what is good and righi It is brought in only in its specific function as reduced complexity as the elimination of other possibilities, since we do not have thi means for calculating ab ovo what are better conditions (Wildavslq and Hammond 1965). Consequently history is accepted only on thi condition that it can be revoked: all planning must indeed work oi the basis of given conditions and needs, but it can change these t< the extent that it can totally replace what exists.42 Hence, all prog ress must take the form of substituting functional equivalents, ex changing given conditions for better ones. The future no longer comes about automatically, but only piece meal by means of decision-making processes. The security whicl once was offered by an undisturbable past must thus be. drawi partially from the future and partially created by guarantees in the status quo. The belief in progress itself becomes an equivalent form of security, needing continual confirmation and finding it in thi quantitative increase of something or other. Planning makes tht future a part of the guaranteed status quo, with the result that «i change of plans must be treated like a change of property.43 To th< individual the future is presented in the form of a career with typ ically expectable stages: he does not absolutely have to reach them but they allow him to determine where he stands.44 It remains an open question whether all this (and much more1 can congeal into a lasting attitude oriented toward a high degret of variability. Only to the extent that it can will social systems, and the political system in particular, be able to hold onto the high degree of complexity they have by virtue of differentiation from tht i TICS AS A SOCIAL SYSTEM 161 'i of society, autonomy, functional specification, and increased -i 11 ctivity. MIL Stabilization of Complexity 1 tie previous analysis can be summarized in the following gen-~ril proposition: the complexity of a political system must corre-*p »'id to the complexity of the surrounding society. This correspondence can be a relation of equality only in the theoretical muting case, where a political or administrative decision corre-N -n nds to each event in society. A demand for such a precise equal-•l\ )f complexity would do away with the differentiation of society -.1 the political system, keeping both systems together at the level *l -imple, archaic relations. The Greek distinction between oikos sm polis had already gone beyond that stage. And even the boldest ■ 1- LS about politicizing all of society, as advocated in the Marxist-í inist tradition or in many developing countries, are not striving i ;qual complexity. Instead, they are taking for granted a sparing an I strategically narrow use of political decision-making power. W. can thus assume that the complexity of the political system is m i than that of society. This situation forces the political system ehave selectively in society, and indeed to develop its own - * 3 of selectivity. The political system compensates for its lesser d _ree of complexity through power. We must presuppose this general relation between environment 3T 1 system in order to grasp adequately what effects an increase nr c omplexity in society has and what problems it poses for the political system. If our basic hypothesis mentioned above is correct, »t can suppose that a significant increase in the complexity of *tK íety (especially through functional differentiation) will alter the «iructures and power relations of the political system, or that struc-fcin s and power relations that are arrested in the political system »ill block such social development. In simpler societies political domination, even when stabilized jsi t learly defined roles, gets by with relatively simple and directly rMiMvated instruments of power—e.g., superior physical force, con- POLITICS AS A SOCIAL SYSTEM 162 trol over part of the economic distribution process, the loyalty of a devoted subgroup, and belief in magical or religious sanctions. Domination is exercised only occasionally in the form of decisions and does not regulate the daily life of the population. If, however, the complexity of society increases (for whatever reasons) and if its need for binding decisions as well as their scope also increases, political-domination can no longer be adequately based on invariant or barely variable constellations of motives. Instead, it must increase its own complexity, making it correspond to the complexity of society. The capacity of those traditional bases of power to be increased is inherently limited. Selections having a low degree of selectivity—orders to drive on the right side of the road, to hand over two cows, to attend an assembly—can be carried out by threats based on force or by appeals to loyalty. The consequences can be surveyed. When, on the other hand, the number of alternative possibilities increases so much that every affirmation excludes un-surveyably many alternatives, the opportunity for achieving a consensus about particular selections becomes extremely small unless powerholders allow themselves to be influenced in the preselection of alternatives. Under these conditions, the traditional instruments of power and the alternatives they permit in the case of noncompliance are shifted to an increasingly marginal position (though not for all that an unimportant one). For normal cases an unquestioning, indeed almost unmotivated, compliance with binding decisions must be guaranteed. That, too, is possible only if the complexity of the political system is great enough, only if the structure of the. political system permits a sufficient number of possible alternatives, and the system thus is free to let itself be influenced. This raises the problem of stabilizing great complexity in the political system. It is the question of how the system can be continually guaranteed to have a sufficiently indeterminate structure and a broad openness to other possibilities.45 A high degree of tolerance for indeterminacy is, as a rule, difficult to institutionalize.46 But that is especially so in this case, since the political system is under pressure to make decisions, to eliminate complexity continually, and to develop effective mechanisms to do this. It has the power to free itself of complexity and is supposed to use it only in a way that maintains complexity at the same time. POLITICS AS A SOCIAL SYSTEM 163 In addition to putting up with structural indeterminacy, the political system must focus on easing structural variation. This leads to destabilizing the structures of political and administrative decision-making—both in law and in setting goals. If we measure the stability of structures by the magnitude of the forces which are in a position to alter them, we can recognize that the result is a lowering of the threshold at which they can be altered. Even rather small forces and normal activities, and not just outbreaks of crisis, will influence the development and alteration of structures. Small causes lead to large effects. It is obvious how dangerous such a destabilization is. It can be combined with order only when there abo exists a guarantee that these small causes are either internal to the system or can be selected by criteria internal to the system. Otherwise any events in society could overthrow structures and lead to chaos. The easing of structural variation can only be introduced in conjunction with a greater autonomy and differentiation of the political system from the rest of society. For theories of historical development, the construction of such an improbable system gives rise to significant difficulties. In no way can we think of seeing it in terms of a continuous and linear evolution, where it would be constructed brick by brick by assembling various territories and rights to authority and domination. The political system takes shape at a more abstract level, requiring reciprocal and circular relations of interdependence, so that establishing and exercising one part assumes that the same has already been done for the other parts and for the whole. This is equally true for the external relations of interdependence that the system sustains with its social environment.47 In addition, in the development from traditional forms of political domination to the modem political system there lies a threshold which, when crossed, leads to a radical change in the structure of the political system. All this suggests a hypothesis which—given our present state of knowledge—can only be formulated speculatively: the complexity of a particular kind of political system cannot necessarily be stabilized at any degree whatever, and thus it cannot develop continuously m a stepwise manner. It depends on too many variables, which can be coordinated only in particular circumstances. Once a certain level of development has been reached, the stability of the previous POLITICS AS A SOCIAL SYSTEM 164 form of political domination disappears. A regression is then likely.48 Only infrequent constellations of factors can help to cross this threshold, since a pronounced differentiation from the rest oJ society, autonomy, functional specification, indeterminacy of structure (especially the positivization of law, the interchangeability oi powerholders, and the mobilization of political support), and significant increases in power are now required. Each of these requirements depends on the meeting of rather complex conditions. As history shows, this step in development is not impossible, but improbable.49 Among other things, it depends on interim solutions remaining sufficiently stable and able to perform the function oi easing the transition.50 In the long run, security can be regained and stabilized without overwhelming tension only if complexity itself becomes a stabilizing factor. In the course of this essay, I have alluded in passing to this possibility in a number of ways—with the idea that the more unstable parts of the system can regulate the more stable ones, and the idea that an opportunistic approach to the satisfaction of values promises stability to the extent that it can fulfill sufficiently many conflicting values by changing preferences and burdens sufficiently quickly. Balancing the dominant path of communication with the reverse path in the power-circuits of the system depends on a high level of complexity's having been distributed to separate situations. The problem of stabilizing indifference and arbitrariness in the relations among subsystems belongs here, along with the problem of learning attitudes toward roles that are more favorable to complexity. The ability to absorb social conflicts also strengthens the system, with the help of complexity. We can summarize all of this in the idea that more complex political systems have a better chance for adapting to an increasingly more complex environment And thus they have better prospects for maintaining their complexitv A theory of the political system that is needed to handle problems of this kind cannot base its conceptual framework on a reconstruction of the traditional, ethical philosophy of political society. Nor can it base its conceptual framework on a formal, political theor> of institutions, focusing for example on various forms and organs of the state. A theory of the political system can do justice to the 1 DLITICS AS A SOCIAL SYSTEM 165 . omplexity of its subject matter only if it reaches a level of theoretical abstraction that.is still not customary today. Systems theory is still unable to offer an elaborate and tested framework for this coal. But it can offer so many relatively consistent suggestions that its further elaboration seems worthwhile.