1
Power as a Communication Medium

Using the theory of communication media as a basis for a theory of power has the advantage of making it possible to draw a comparison between power and other communication media, applying identically composed questions, comparing it, for example, with truth, or with money. Thus, these questions not only serve to clarify the phenomenon of power, but at the same time help produce a more broadly orientated comparative interest and facilitate the exchange of theoretical insights between different media areas. In addition to such new insights, the theory of power gains a general perspective over forms of influence which will be discussed separately from a more designated concept of power. This makes it possible to avoid what has so often been remarked on, i.e. overloading the concept of power with attributes of a very broadly and loosely defined process of influence.1

By way of introduction, therefore, a few cursory remarks about the theory of communication media are necessary.²

1. Societal theory is, according to the main elements it has inherited from the nineteenth century, on the one hand a theory of social differentiation into strata and into functional subsystems and, on the other hand, a theory of socio-cultural evolution. Both aspects are intertwined through the assumption that socio-cultural evolution leads to increasing differentiation. Within this frame of reference, issues of communication and questions of motivation for accepting and complying with communications remain incompletely illuminated. Partly they were seen as merely psychological actions and attributed to the individual, so that they could be passed over in a macro-sociological approach. Partly they were subsumed under such special concepts as consensus, legitimacy, informal organization, mass communication, and the like. Both ways of dealing with the problem led to concepts of a lesser order and of more limited range in comparison with the concepts of differentiation and evolution. Questions of communication and motivation were thus not entirely excluded from societal theory but they did not rank with the main concepts. Against this one could speak out in the name of a supposed humanism and deplore the loss of humanity without achieving any more than articulating protest at a quite inappropriate level.³

The attempt to formulate a general theory of symbolically generalized communication, and to tie it in with the concept of societal differentiation as well as with statements about mechanisms and phases of socio-cultural evolution, is aimed at filling this gap. In this we are aiming to avoid recourse to the ‘subject’, in the way that transcendental philosophy has used the term, as well as any claim to deal with the physiologically and psychologically concrete individual. The one alternative would be too abstract, the other too concrete for sociological theory.⁴ Instead, we shall proceed from the basic assumption that social systems are only ever formed through communication, i.e. they always assume that multiple selection processes determine one another by anticipation or reaction. Social systems first arise from the need for agreed selections, just as, on the other hand, such needs are first experienced in social systems. The conditions which make this correlation possible are the result of evolution, and change alongside it. Just as evolution articulates the temporal, and differentiation articulates the material, so communication articulates the social dimension of the societal system.

Communication comes about only if one understands the selectivity of an utterance, and that means the ability of the system to use as a selection one of the conditions which allow for its own existence.⁵ This implies contingency on both sides, and thus also the possibility of rejecting the selections on offer transmitted by the communication. These possibilities of rebuttal, as possibilities, cannot be eliminated. A rejection communicated in return and the application to subject areas of that rejection within social systems constitutes conflict. All social systems are potential conflicts; it is only that the degree to which this conflict potential is realized varies according to the degree of system differentiation, and according to societal evolution.

Under such constitutive terms, the choice between ‘Yes’ and ‘No’ cannot be guided by language alone, for it is precisely language which guarantees both possibilities; neither can it be left to chance. Therefore, in every society there are devices additional to language which guarantee the transmission of selections to the appropriate degree. The need for these devices increases and their form changes with the evolution of the societal system. In simple societies these functions are mainly fulfilled by ‘reality constructions’ founded on shared, lifeworld experiences, which underlie the communication processes and are taken for granted.⁶ To a great extent language serves to confirm such assumptions, and its potential for negation and information does not become exhausted.⁷ Only more advanced societies develop the need for a functional differentiation between the language-code in general and special, symbolically generalized communication media, such as power or truth, which condition and regulate in particular the motivation for accepting offered selections. By means of this differentiation, potentials for conflict and consensus in society can be mutually extended. The evolutionary mechanisms for the variation and selection of usable, socially effective, transferable selections part company and this speeds up socio-cultural evolution, since new choices can be made under more possibilities from more specific points of view.

The invention and spread of writing seem historically to have been the especial cause of the development of symbolic communication media. Writing greatly widened the potential for communication in society beyond the interaction of people immediately present and thus removed it from control by concrete interaction systems.⁸ Without writing it is impossible to create complex power chains in political and administrative bureaucracies, let alone democratic control over political power. Ostracism requires writing. The same applies to the discursive development and perpetuation of more complex elaborations of truth statements.⁹ The classifying function of a logically schematized truth-code is needed only when a body of thought formulated in writing is available. But even the moral generalization of a special code for friendship/love (philía, amicitia) in the Greek polis is a reaction to the written culture of the city, a compensation for a density of interaction among neighbours (philói) which can no longer be assumed. Above all, this dependence on writing is evident in the money-code. It is only through the second coding of language through writing that the societal communication process is released from the bonds of social situations and unproblematical assumptions to the extent that, in order to motivate the acceptance of communications, special codes have to be created which, at the same time, also condition what can successfully be maintained and assumed.

2. Accordingly, communication media shall be defined as a device additional to language, namely, as a code of generalized symbols which guides the transmission of selections. Language normally guarantees intersubjective comprehension, i.e. the recognition of the selection of the other party as a selection. Communication media add to this by providing a motivating function in that they urge the acceptance of other people’s selections and as a rule make that acceptance the object of expectations. Accordingly, communication media can always be formulated when the manner of selection of one partner serves simultaneously as a motivating structure for the other. The symbols of this connection between selection and motivation then take on the function of a transmission and make clear the connection between the two aspects, so that this anticipatory connection can strengthen and, in addition, motivate the selectivity.

This concept contains a number of assumptions and implications which also apply to the theory of power and steer it in a particular direction. The first and most important assumption is that media-guided communication processes bind partners who both complete their own selections and know about this from each other.¹⁰ Let us use the terms ‘alter’ and ‘ego’. All communication media assume social situations with the possibility of choice on both sides, in other words, situations of double contingent selectivity. That is precisely what gives these media their function of transmitting selections from alter to ego while preserving their selectivity. To this extent, the initial problem in all symbolically generalized communication media is the same; what applies to love or truth applies to power. In each case the influential communication relates to a partner who is to be directed in the making of their selections.¹¹

According to this, transmission of selection outputs means precisely the reproduction of those outputs in simplified form abstracted from the requirements of their initial configuration. It is precisely in view of this simplification and abstraction that symbols are needed to replace the concrete beginning, the initial link in the selection chain. For this purpose communication media develop symbolically generalized codes which then act as shared orientations for their operations. Yet each subsequent phase of the process continues itself to be self-selection. Communication media consequently provide the ability to orient around widely different situations while, at the same time, enabling particular (non-identical) selections to be made. It is only under this basic condition that power, too, functions as a communication medium.¹² It organizes social situations through double selectivity. Therefore, the selectivity of alter must be distinguished from that of ego, for very different problems are raised in relation to their respective positions, particularly in the case of power.

Accordingly, a fundamental assumption of all power is that uncertainty exists in relation to the selection of the power-holding alter.¹³ For whatever reasons, alter has at his disposal more than one alternative. He can produce and remove uncertainty in his partner when he exercises his choice. This deviation via the production and reduction of uncertainty is an absolute precondition of power. It determines the latitude which exists for the generalization and specification of a particular communication medium – and is not, for instance, one particular source of power among others.

Power also assumes openness to other possible actions on the part of the ego affected by power. Power invokes its performance of transmitting by being able to influence the selection of actions (or inactions) in the face of other possibilities. Power is greater if it is able to assert itself even in the face of attractive alternatives for action or inaction. And it can be increased only in combination with an increase in freedom on the part of whoever is subjected to power.

Power must therefore be distinguished from coercion (Zwang) to do something concrete and specified. The possible choices of the person being coerced are reduced to zero. In borderline cases coercion resorts to the use of physical violence and thereby to the substitution of one’s own action for unattainable action from others.¹⁴ Power loses its function of bridging double contingency the closer it comes to coercion. Coercion means the surrender of the advantages of symbolic generalization and of steering the partner’s selectivity. The person exercising coercion must himself take over the burden of selection and decision to the same degree as coercion is being exercised – for many cases we can even say that coercion has to be exercised, where there is a lack of power. The reduction of complexity is not distributed but is transferred to the person using coercion. Whether this is the sensible thing to do depends on how complex and variable are the situations where decisions about action have to be made.

The use of coercion itself can only be centralized in very simple societal systems. More complex systems can only centralize decisions (or even decisions about deciding the premises for making decisions) about the use of force. That means that they must develop power to make coercion possible. The concept of a ‘coercive apparatus’ introduced by Max Weber covers this situation.

Even these simple initial reflections show that a closer definition, operationalization and measurement of concrete power relationships is an extraordinarily complex enterprise. A multidimensional measure would have to be used to assess the complexity of the possibilities from which both sides (or, in chain foundations, all participants) can choose an action.¹⁵ The power of the power-holder is greater if he can choose from more and more diverse types of decisions for power-like assertions and it is greater if he can do this in opposition to a partner who himself has more and more diverse alternatives. Power increases with freedom on both sides, and, for example, in any given society, in proportion to the alternatives that society creates.

According to this, not only scientific and methodological problems are indicated.¹⁶ Further, the result of this complication for society itself is that it must develop substitutes for an exact comparison of power constellations and that these substitutes themselves become a factor in power. Firstly, hierarchies which postulate an asymmetrical distribution of power serve as substitutes. It is assumed that a superior has more power than an inferior (although in bureaucratic organizations the opposite may be normal).¹⁷ Another substitute is the history of the system; cases of past successes in obtaining desired outcomes in conflict situations are recalled, normalized, and generalized as expectations. This function, as a basis for comparison, goes together with the symbolic explosiveness of concerns about status and single events, which are used in a way that brings present power constellations too much to the foreground. Thirdly, there are important possibilities for substitution in semi-contractual arrangements, whereby a too-powerful partner comes to terms with those who might withdraw or be disloyal.¹⁸ In all these cases the direct communicative recourse to power is replaced by reference to symbols which commit both sides normatively, and at the same time take into account the presumed power differential.

These are all functional equivalents for measuring power and for tests of power as decision premises in societal reality. The institutional anchoring and practicability of such substitutes renders exact calculations unnecessary, and even makes any attempt to do so problematic. The result of this is that if science were to produce a way of measuring power, it would alter social reality; that is to say, it would destroy the substitutes and reveal them as false assumptions. More within the bounds of probability, however, is that science will develop its own substitutes for measuring power, which would be dealt with in other areas of society, as purely and simply the province of science.

3. The function of a communication medium lies in transmitting reduced complexity. The selection made by alter limits ego’s possible selections by its being communicated under particular and specified conditions. Dependencies that pass through communication media are distinguished from general interference and mutual hindrance (such as alter listening to the radio and ego being unable to go to sleep) in that they presume some process of communication which can be pre-conditioned by symbols. Thus they are subject to cultural formation, can be changed by evolution, and are compatible with a large number of system conditions.

Also, in the case of power, it is this transmission of the results of selection which is the main point of interest, not, for example, the concrete causes of certain effects. Power does not arise only in the borderline case where alter concretely lays down ego’s action, for instance to insert a given screw as tightly as possible. It is more typical, and satisfactory, to view power, in the same way as every other communication medium, as putting limits on the partner’s range of selections.¹⁹ The causal notion hitherto guiding theories of power²⁰ is not to be dismissed but is abstracted. It does not designate an invariable link between concrete conditions in the world – expressions of power, and behaviour – nor does it restrict the effectiveness of power to the case where the behaviour of ego would have taken a different course without the power-transmitting communication from alter.²¹ If such were the case, it would be assumed, wrongly, that there always exists a ready resolve (that can be empirically determined), by now in existence, which is then broken. In fact, however, the existence of a power differential and the anticipation of a power-based decision make it quite senseless for the subordinate even to make up his mind. And it is in precisely this that the function of power lies. It secures possible chains of effect independent of the will of the participant who is subjected to power – whether he wishes it or not. The causality of power lies in neutralizing the will, but not necessarily in breaking the will of the inferior. It also affects him, and that most precisely, when he intended to do that same thing and then learns that he has to do it anyway. The function of power lies in the regulation of contingency. As with every other media-code, the power-code relates to a possible (!) and not necessarily an actual, discrepancy between the selections of alter and ego by creating the impression of equality between the two and so removing this assumed discrepancy.

Accordingly, the power of the power-holder is not satisfactorily described as a cause, or even a potential cause. It can be compared rather with the complex function of a catalyst. Catalysts accelerate (or decelerate) the triggering of events; without themselves changing in the process, they cause changes in the ratio of effective connections (or probability) expected from chance connections between system and environment. Thus, in the end they produce a gain in time – always a critical factor for the construction of complex systems. In this respect – and this we will term, following Kant, the concept of schematism – they are more general than their respective products. In the process of catalysis, they do not change, or do not change to the same extent as the accelerated (or decelerated) process produces or inhibits effects.

Bearing in mind that we are here talking about a real structure (and not only an analytical summary),²² one can then say that power is an opportunity to increase the probability of realizing improbable selection combinations.²³ Real probabilities contain a tendency to be selfreinforcing. If one knows that something is probable, one prefers to reckon that the event will happen, rather than that it will not, and the more relevant it is, the lower the threshold which starts such a process moving. The same applies to improbabilities – as every driver knows. Therefore, a prior decision is necessary each time to decide whether to view an uncertain event as (very/quite/not very) probable or as (not very/quite/very) improbable. Here purely psychological laws can play a part.²⁴ In addition, social definitions of the situation will come into play and influence the perception of the probable or improbable. And for their part these situational definitions can be presented by means of symbolically generalized communication media and be conveyed as general formulas. The catalyst function of power is, therefore, based on very intricate causal complexes. This is precisely why power is to be understood only as a symbolically generalized communication medium. The development of abstract formulations by way of symbolically controlled selection complexes ensures at the same time that power is not seen as something dependent on direct action and interference by the power-holder on the power-subject.²⁵ It is only by assuming a process of communication, that is, that the power-subject always learns by some indirect route of the selectivity (not only of the existence!)²⁶ of past or future power-acts by the power-holder. It is especially a function of generalizing the communication medium of power to make possible such deviations without thereby cancelling out the identification of the power-code and the topics of communication.

4. It is typical of all communication media that at the base of their differentiation lies a special interaction constellation and, within this framework, a specific fundamental problem. Communication media are raised above the taken-for-granted reality of normal life only where influence is contingent and thereby, for once, rather improbable. Only when, and in so far as, goods are scarce does the active claim to some of them on the part of one person become a problem for others, and this situation is then regulated by a communication medium, which transfers the action selected by the one person into the experience of the others and so makes it acceptable.²⁷ In the context of scarcity, influence becomes precarious in a special way, so that, in view of this unusual situation, a specifically generalized communication medium can take shape, which makes possible the transmission of reduced complexity in this case, but not in others. This is also how truth arises. Here too, with the framework of unproblematical assumptions and beliefs, a certain improbability of information must first arise before test criteria start to function, and before a special code can be formed to regulate the identification of truth and untruth. Truth is doubt overcome. It can be triggered by the simple disappointment of cognitive expectations, but also by a set of cognitive instruments with sharply increased capacity for resolving the issue.

Such a focus, a passage through increased contingency, is also necessary for the formation of the communication medium of power. Not every execution of a proposed action becomes problematical. One does not let go of something one has been given, but accepts it and holds it tight, etc. But in special cases, if the proposer, shall we say, restricts himself to proposing, and his own action restricts itself to prescribing the action of others, the actual context of circumstances can no longer perform all the transmission of selections which are required. The temptation to negation also increases with the contingency of the selection. Then transmission of selection can occur only under special presuppositions and the power-code reconstructs and institutionalizes these presuppositions. Only with the help of a symbolically generalized communication medium do they become the basis of reliable expectations.

It is difficult to encapsulate this problem in a definition which states categorically what power is and is not. The terms of the problem, however, evoke distinctive and describable sets of circumstances. One can state that the greater the extent to which influence becomes contingent, on account of it being recognized as an action whose selectivity refers only to the activation and guidance of someone else’s action, the less the extent to which a natural-situational congruence of interest can be subsumed; that the more problematical the motivation becomes and the more necessary a code which regulates the conditions of transmission of selection and the attribution of personal motives. This approach, which proceeds from interaction configurations, can then be taken over into the theory of societal evolution in the thesis that, as societal differentiation increases, so does the frequency of situations in which, no matter how high the degree of contingency and specialization, transmission of selection must take place if an achieved level of development is to be maintained. In important functional areas, situational congruence of interests no longer occurs frequently enough or with enough specialization for this to suffice. Then the development of a special code for power, tailored to such problems, becomes an unavoidable priority for further evolution.

This line of argument, moreover, has its parallels in other media areas and is supported by them. Only from a certain stage of development onwards does daily communication become so loaded with information that truth itself becomes a problem. Only from a certain stage of development does the stock of goods become so large that it becomes meaningful to keep it open to contingent intervention in situations of scarcity. It could further be said that love becomes necessary as a special communication code only when others’ emotions and images of the world are so strongly individualized – and that means, have become so contingent – that one can no longer be sure of them, and therefore, according to cultural standards, one has to love for its own sake. And even art, as a communication medium, depends on increased contingency, i.e. the contingency of works manifestly produced, but no longer sustained by the context of specific, lifeworld purposes. Designated in all of these are problematic areas of interaction, namely variants of the problem of selection transmission and, at the same time, stages in the evolution of the societal system.

5. Perhaps the most important difference as against older theories of power is that the theory of communication media conceptualizes the phenomenon of power on the basis of a difference between the code and the communication process and is therefore not in a position to attribute power to one of the partners as a property or as a faculty.²⁸ Power ‘is’ code-guided communication. Attribution of power to the power-holder is regulated in this code with wide-ranging results concerning the reinforcement of motivations to comply, responsibility, institutionalization, giving specific direction to wishes for change, etc. Although both sides are acting, whatever happens is attributed solely to the power-holder.²⁹ Scientific analysis, however, should not let itself be side-tracked by rules of attribution which are contained in their object. Such rules do not cause the power-holder to be more important or, in any sense, ‘more causal’ in the formation of power than the power-subject.³⁰ The rules of attribution contained in the media-code are themselves another possible object of scientific analysis.³¹ One can, yet again, also raise questions about their functions. To this end, the analytical apparatus must first abstract predecisions from attributions. At the same time this demands a greater differentiation of society’s scientific subsystem, in our case a more far-reaching differentiation between science and politics.

The difference between the generalized code and the selective communication process will constantly be with us in what follows. The symbolic generalization of a code, according to which expectations can be formed, is a prerequisite for the differentiation of power as a specialized medium capable of being related to particular combinations of problems; it produces certain actions and is subject to certain conditions. Further, the generalized media-code contains the starting points for cumulative development in the course of societal evolution. From these points of view, power is of interest for the theory of society. This should not exclude the possibility that theories of organization and interaction could work with simplified concepts of power, such as those which already presume in their conception of power differences of status or sufficient possibilities for information and calculation. However, it would be impossible within the frame of such circumscribed premises to reach any conclusion about the implications of power for wider society.

6. In a much acclaimed, extensive critique of the work of Parsons, and of his power theory in particular, Alvin Gouldner expresses his surprise that Parsons, in his treatment of power as a symbolically generalized medium, identifies it so closely with legitimate power, with ‘establishment power’, and that he takes this as normal for society.³² Pointing to the brutality and selfishness of power-holders, he dismisses this view, across the board as well as in its individual formulations, as unrealistic, as intellectually absurd, as utopian, as misleading. This astonishment on the part of a sociologist is itself astonishing to sociologists, even more so because it is formulated in the framework of a sociology of sociology. Of course, it is indisputable that sociology can and should concern itself with the phenomena of the brutal and selfish exercises of power. Such an interest, however, should not grow into a prejudiced view concerning social reality, built into concepts and theories.

The real achievement of Parsons’ theory was to replace the prejudices of sociology as a science of crisis and opposition with a relatively non-issue-driven theoretical design (one thus open in its turn to criticism). However one judges the adequacy of this analytical apparatus, it is indisputable that the institutionalization of enforceable legitimate power is a phenomenon of greater social significance in comparison with brutality and selfishness. Everyday social life is determined to a much greater extent by recourse to normalized power, i.e. legal power, than by the brutal and selfish exercise of power. Exceptions limited to certain areas actually serve to illuminate this state of affairs.³³ Intervention by legitimate force is more considerable; one simply cannot think it away without disrupting and transforming almost all normal social life. Brutality and selfishness are phenomena which are compatible with many social conditions so long as they do not undermine the dominance of institutionalized power. Such an argument, of course, does not justify any single brutal act, and, moreover, does not justify tolerating or accepting it, as one knows from the history of theodicies³⁴ and welfare. But this kind of accounting problem is really secondary – both historically and theoretically. It presumes the introduction of a binary schematism to differentiate debit from credit, or right from wrong, or conformity from non-conformity.

By working out a theory of symbolically generalized communication media, we are trying to avoid such controversy. The conditions for forming a dichotomy between ‘ruling order’ and ‘critique’ are part of the theory itself. The theory treats such disjunctions as elements of a communication code and asks about their genetic preconditions, their functions, their results, their complementary mechanisms, their chances of development. Such a theory can also be characterized, as Gouldner would have it, as moralistic and conservative, if one assumes that the theory confirms the social characteristics that it has exposed. It is conservative in the sense that it wishes to retain and keep open the option of expressing an opinion either for or against a manifestation of power according to prevailing circumstances.

Notes

1 Social psychologists in particular run this risk. Typical examples would be: Raven (1965), and Clark (1965).

2 The treatment of power specifically as a communication medium begins with Parsons (1963a). For other suggestions, applications and criticisms, see Chazel (1964); Mitchell (1967); Lessnoff (1968); Giddens (1968); Turner (1968); also Baldwin (1971a) and Blain (1971). In what follows the concept of the communication medium is used in a way which is independent of Parsons’ interchange paradigm; it is therefore not built up on an idea of exchange, and also differs in other respects from Parsons’ concept. The differences depend on the interpretation of the contingency problem (see Luhmann (1976)).

3 Cf. for example Homans (1964) and Maciejewski (1972).

4 This statement is meant at the same time to show how problematic it is to designate the individual as subject. With such equivocations it becomes all too easy to step from the abstract into the concrete.

5 For this concept of communication, cf. MacKay (1969).

6 Cf. Berger and Luckmann (1969); McLeod and Chaffee (1972); also Arbeitsgruppe Bielefelder Soziologen (1973).

7 See, for example, Marshall (1961).

8 On this cf. Goody and Watt (1963) and Goody (1973).

9 ‘Dialogue’ is then cultivated as a literary form, as a self-contradictory protest against the demands of literacy; and only in this way does it achieve stylistic perfection.

10 Parsons accommodates this by using the idea of a double contingency as a prerequisite for the formation of complementary expectations. Cf. Parsons and Shils (1951), pp. 14ff.; see also the remarkable definition of authority as the laying down of decision-making premises (not decisions!) of others in March and Simon (1958), p. 90.

11 Accordingly, in the context of a general power theory, it is not really sensible to give a one-sided emphasis to the decision-making problems of any one side. In particular power areas this may well be different. Thus Fisher (1969) recommends that offices which deal with foreign affairs concern themselves less with making their own policies more precise than with paying attention to other states and above all working out what decisions from other governments would be acceptable before any exercise of power.

12 Abramson et al. (1958) in particular emphasize that power theory must take into account the majority of possible actions by both sides.

13 Cf. Crozier’s observation (1963), especially pp. 193ff., that in heavily structured organizations power shifts to where, in relation to choices of action on which others depend, there is still a remnant of uncertainty. For an elaboration on a general ‘strategic contingency theory’ of power see Pennings et al. (1969); Hinings et al. (1974).

14 This case of using physical force – moving different bodies in order, for example, to cause them to change their position in space – must be carefully differentiated from the symbolic use of physical force in order to form power. We will return to this in Chapter 4.

15 One problem, which can only be noted here, is that all these measures are relative to the conditions of possibility which one takes as a basis. Measurement thus always takes for granted that participants belong to one system and are restrained by common conditions of the possible.

16 See Danzger (1964), pp. 714ff.

17 See, for example, Walter (1966) and a criticism of this: Mayhew and Gray (1969). See also pp. 117ff.

18 Cf. on this a series of experiments which proceed specifically from alternatives open to both sides: Thibaut and Faucheux (1965); Thibaut (1968); and Thibaut and Gruder (1969).

19 This is sometimes taken into account explicitly in power theories, but is usually implicit. For a corresponding definition of the concept of power, see van Doorn (1962/3), p. 12.

20 A more recent survey of problems in a causal theory of power can be found in Dahl (1968), pp. 46ff. Cf. also Gamson (1968), pp. 59ff.; also Stinchcombe (1968), pp. 163ff., who suggests an information theory formulation of the causal concept of power.

21 This characteristic is frequently taken account of in the form suggested by Max Weber. One assumes power only where the power-holder can assert himself even against opposition (see Weber 1948, p. 28). Emerson (1962) and Holm (1969), for example, use this conceptualization. Initially only the selective definition, the ‘processing of contingency’, is important on the level of a general theory of communication media. We will return to the specific characteristics which this process takes on in the case of power.

22 This conception goes back to Max Weber’s concept of ‘chance’. Wrong (1968), pp. 677f. points out quite correctly that the assessment of the power-subject is being referred to and not the statistical analysis by sociologists of cases of the actual exercise of power.

23 In this sense Dahl (1957) formulates not only power itself as ‘chances’ but also the causality of power as a change of probabilities.

24 For example, a generally observable tendency to prefer positive situation definitions, which probably only seldom lets the case of double negation (less improbable) occur. Cf. for example Jordan (1965); Kanouse and Hanson (1971). Another question would be whether negative or positive verdicts possess greater interference with alternatives, whether the improbable or the probable block or leave open more varied possibilities.

25 Both for and against this see Nagel (1968), pp. 132f.; Gamson (1968), pp. 69f.; and Wrong (1968), pp. 678f.; also Schennerhorn (1961), pp. 95f., using the example of the local political power of large corporations.

26 Thus Nagel (1968).

27 See Luhmann (1972a).

28 This redistribution has an even more decisive effect in the context of other communication media. Truth, seen as a communication medium, can no longer be characterized as a quality of ideas or sentences, love no longer as feeling, money no longer as possession, belief no longer as an inner binding of the person. For the sociology of communication media, such ideas and attributions do not characterize the theory but its object: simplifying aids to understanding social life orientated to generalized codes.

29 As we know from experimental research, even hierarchies in general guide the process of attribution in this sense. Cf. Thibaut and Riecken (1955).

30 The reverse of course applies equally little, that is, viewing the power-subject as the decisive cause of the formation of power. Thus numerous American definitions of authority with reference to Barnard (1938), pp. 161ff.; Simon (1957); Peabody (1964); also, concerning threats, Lazarus (1968), pp. 339ff.; Fisher (1969).

31 Lehman (1969), for example, points out the increased significance of a stable, predictable imputation of power on the macro-sociological level. I do not know of any more detailed examinations.

32 See Gouldner (1971), especially pp. 290ff.

33 Cf. Guzmán et al. (1962).

34 Editors’ note: Theories of why God allows evil to exist in the world.