Chapter 3
If the last chapter focused on the production of knowledge, this one is primarily concerned with its distribution. Taken together, the two chapters lay the groundwork for a political economy of information.
All communication is costly and hence selective. Communication, to be effective, must overcome technical, semantic, and pragmatic barriers to the transmission of information. Meaningful communication always requires some minimum sharing of context between sender and receiver. Where this is difficult one must resort to shared abstractions in order to be understood. Abstraction facilitates the diffusion of a message since it increases the number of particular situations in which a message can have utility and relevance.
The U-space (U stands for ‘utility’) explicitly links the diffusibility of a message to its degree of abstraction. Trajectories in the U-space towards well-diffused and abstract knowledge are affected by the prior distribution of knowledge in the space as well as by the social power of those in possession of such knowledge. Both are stabilized by institutions, structures designed to economize on social information processing as well as to minimize the social production of entropy.
Coding, or codification as we shall henceforth call it, also facilitates diffusion. This is indicated in the C-space (C stands for ‘culture’), where different kinds of knowledge can be represented.
Effective communication requires the sharing of either abstract codes or contexts between sender and receiver. The choice is affected by the communication technologies available. The data field emerges from a sequence of such choices, a product of the way that the forces of codification, abstraction, and diffusion interact. The data field exhibits self-organizing properties that challenge the conventional economic view of information as a ubiquitous and static phenomenon.
Cognition as a social process
In the preceding chapter, we examined data processing as a phenomenon occurring within a single brain, a highly dense and connected network of neurons capable of emergent responses to patterns of incoming stimuli. In this chapter we are less concerned with connections within brains than with connections between them; the focus is thus on the communicative component of data processing now taken as a social activity. Does the connectionist perspective, however reformulated, have something to offer at this level and what kind of emergent phenomena might it give rise to?
We can only partly address such questions in what follows and we do so very indirectly, relying on later chapters to flesh out our reply. Our emphasis will be less on the production of knowledge – the subject of the previous chapter – than with its distribution, with how and with whom it is shared once it has been produced. We cannot, of course, entirely ignore the production of knowledge since the prospect of having to share it with others will usually influence how it gets articulated and how it subsequently evolves.
The production and distribution of wealth in society have traditionally been the concern of political economy. If we take information to be the dominant form of wealth in most post-industrial societies – that is to say, if we treat it as an object of, as much as a support for, economic exchange – we can view our mission in the preceding chapter and in this one as preparing the ground for a political economy of information. The production of information results from a complex interplay of data coding and abstraction;1 its distribution, as it turns out, activates the same variables. Does this open up the prospects of developing a unified scheme?2
Transmitting tacit knowledge
How can the experience of a concert be properly shared with someone who was not present? How can riding a bicycle be taught except by force of example? What exactly is it about someone's facial expression that signals to intimate friends that he/she is in a bad mood? All these questions butt up against what Polanyi has called ‘tacit knowing’:3 large tracts of conscious experience and behaviour located in the south-east region of the E-space, as we have already seen, habitually escape verbal formulation or coding. Few genuinely creative people, for instance, are able to articulate in a convincing way the processes that lead from initial inspiration to the execution of their work. For this reason, Picasso, as a painter, is often better understood by reading his critics and interpreters than by perusing his own pronouncements. Even then, there will remain a large uncodifiable residue in his art that can never be completely grasped. This residual Picasso will remain the object of continued speculation and successive reinterpretations which are more likely to reveal something of their authors as they grapple with the unfathomable, than anything objectively ascertainable concerning the artist himself.4
Tacit knowing describes the form in which we hold our least communicable knowledge assets in the E-space. It was to these that the early Wittgenstein referred to in the Tractatus when he asserted that ‘whereof one cannot speak, thereof one must be silent’.5 Yet speaking up or shutting up are unnecessarily stark choices6 given the potential richness of our communication resources. After all, a painting by a Picasso or a Jackson Pollock remains an intentional act of communication even if what is actually externalized in paint is weakly coded and can only partially convey what is experienced by the artist at the moment of executing the work. Communication between the artist and his audience, however, is only possible in so far as some overlap exists between their respective E-spaces. And whether in fact communication occurs will depend on how well the artist successfully reconciles his divergent needs for autonomy in the processing of personal experiences on the one hand, and for sharing these with a sympathetic audience on the other; a balance must be struck between the claims of self-expression and those of effective communication.
Individual cognitive and affective processes and communication needs thus exert a reciprocal influence upon each other so that the way that we know something affects the way that we express it and, conversely, a concern to express something affects the way that we know it.
The cost of communicating
In his book Personal Knowledge, Polanyi's interest is centred almost exclusively on the individual knowing subject. Although he pleads in favour of a social system that can restore to individual (personal) knowledge its sense of wholeness through an acknowledgement of its tacit component, having stated its requirements, he then tends to leave it at that. He does not go on to ask what kind of social system is likely to do this or what influence it might exert on this tacit component.
In this chapter we explore social knowing, moving from personal knowledge individually held to personal knowledge collectively held, and from thence in Chapters 5 and 6 to differences in forms of social organization and cultures that express variations in the distribution of knowledge. Our guiding thoughts will be the following: communication, like data processing, is an entropy-generating phenomenon that will be subject to economizing efforts.7 The marginal cost of communicating from certain regions of the E-space, for example, will be lower than from others and, other things being equal, these will be the regions preferred for communicative exchanges. Yet such a cost – largely a matter of communicants having built up shared codes that are then available to them in that region – is only part of the story: to properly exploit the value of the message communicated (i.e., its perceived relevance to recipients) will also call for a prior investment in the build-up of shared knowledge assets in the region, assets that can then create a suitable context for the subsequent interpretation of the message. The expected value of the message establishes its terms of trade, so to speak, and hence the productivity of any exchange relationship associated with it. Where perceived message value is high, one will be willing to incur greater communication costs than where it is low, and in recurrent situations one may additionally invest in data processing activities in order to shift knowledge assets into the regions of one's E-space from which effective communication is achievable at lower cost.
From this perspective, auto-communication, the processing and transmitting of data internally within the individual organism, is always a moment in a wider communicative act as well as at times a preparation for it. The requirement of social exchange, therefore, and the benefits that can be derived therefrom, must be counted among the key forces that drive the evolution of knowledge in the E-space. However these are measured, it is often the expected returns to one's communicative efforts rather than to one's data processing activities alone that compel us to shift at least some of our cognitive assets into the north-east corner of the E-space from where they can most easily be transmitted.
Freedom to choose: communication as selection
An ability to communicate is intimately related to an ability to abstract and to code. An experience that is refractory to coding and that cannot be fitted into any category or system of categories is, in the final analysis, incommunicable. To quote Bruner: ‘If perceptual experience is ever had raw, that is, free of categorical identity, it is doomed to be a gem, serene, locked in the silence of private experience.’8 Yet such gems turn out to be constituent elements of all experiencing, in effect a residue of data and information sacrificed through the selective action of the coding process itself. It is in the paradoxical nature of all acts of communication that they must forever remain partial in order to be effective. Pace Wittgenstein, the ability to talk can thus never be wholly disentangled from the willingness to stay silent.
Silence, as a component of the communication process, holds more appeal for some than for others. It may appeal more, for example, to individuals whose personal cognitive or learning style locates them in a part of the E-space where their experience is hard to articulate and hence costly to communicate. We must therefore expect considerable variation in the effort individuals are willing to invest in pushing what they know tacitly towards greater articulation and generality. For much will depend on where in the E-space they set out from; on what cognitive risks and costs they incur in trying to shift, even if temporarily, from one way of knowing to another; and on what returns they can expect for their pains.
The somewhat utilitarian mental calculus just described robs the communicative process of its simplicity and its directness. In particular it poses something of a challenge for those who believe that an individual's mental processes can be inferred from nothing more than a knowledge of his/her material situation. To a cultural materialist like Marvin Harris, for example,9 the high degree of dependence of W3 products and W2 processes on the physical constraints imposed by W1 robs the ‘private language’ of situated individuals, and indeed the private – i.e., emic – language of a situated group of any ability to achieve any ‘real’ or objective understanding of their respective situations.10 Only the language of external observers can do this using ‘etic’ description.
We shall have an opportunity to explore the cultural materialist position further in Chapter 5. Here we must merely note that while such materialism does not go as far as behaviourism in denying ‘mind’ any reality at all, by making its workings overly dependent on W1 it saps it of any emergent properties and hence of its scope as an autonomous level of explanation for an account both of the evolution of W3 and of the way that W3 objects embed themselves in W1. Yet any effort at representation – whether for internal use or for external transmission and whether emic or etic – confronts us with different coding possibilities and choices, some being more parsimonious than others with respect to the amount of data to be processed; how we choose to represent events to ourselves, for example, may not correspond to how we might describe them to intimate friends or to how we would communicate them to strangers. The materialist is committed to the view that these choices, being grounded in a single objective physical reality, are highly constrained in the way that they can legitimately represent it. He holds, in effect, a ‘correspondence theory of truth’11 in which representations that diverge from the templates provided by objective material conditions in W1 are treated as so much ‘superstructure’.
Communicative strategies
Coding choices have to satisfy both internal representational and transmission needs which can only themselves be formulated through further representations. To an outside observer, according to Dennett, someone involved in making such coding choices has to be treated as an intentional system, one whose behaviour can usually be explained by ascribing to it beliefs, desires, hopes, hunches, fears, intentions, etc.12 Does she wish vaguely to connote something in a poetically elliptical fashion only intelligible to a restricted circle of illuminati, or does she aim to denote it with the precision of an American contract lawyer? It may not always serve one's best interests to be clear with everyone. Argyle13 has spoken of the valuable property of vagueness whereby interactors are not committed to a particular relationship and shifts in attitude can be made quite easily. Much body language has this property. Skill in communication, then, can be described as the ability to select positions on the coding and abstraction scales that are appropriate to the circumstances and knowing when to switch from one position to the other. It is a skill that helps each individual to establish, within the limits of what he can articulate, the degree of permeability to outsiders of his personal E-space.
Not all positions that might be chosen along the two scales, however, are equally well endowed in readily usable codes. In the lower section of the coding scale, for example, the codes are harder to articulate; they are less numerous and more ambiguous, and thus can at best meet the needs of internal representation rather than external communication. And at the zero point of the scale, arguably, no codes exist at all; as one moves up the coding scale, however, the codes available increase as do the number of communication channels within which they can be deployed. Furthermore, channel combinations greatly extend the complexity and subtlety of messages that can be sent. When several channels are used in combination for the transmission of a given message, however, by increasing the coding choices available relative to the length of the message they bring us once more down the coding scale and thus effectively reduce the message's degree of coding.
In the upper segment of the coding scale, message data has to be highly compressed to exploit the specialized codes and artificial communication channels that may be available. Unlike the sensory channels that we use, almost unconsciously at times, to communicate with our immediate environment, artificial channels are created by technology. Their spatial reach may be greater than that of sensory channels but they are more costly to develop and use. To transpose a message from a sensory channel to an artificial one may require a non-intuitive effort to adapt and translate it and may call for the mastery of a new coding repertoire.
Here one's communicative strategy indeed reflects material conditions – i.e., the technical choices available – but only in the trivial sense that without a conscious effort at coding no communication at all takes place. Material conditions in such circumstances might therefore constrain communication possibilities but they do not necessarily dictate them; indeed, with technological evolution, communicative constraints are being decreased in ways that progressively extend our freedom of choice. Thus the cognitive autonomy of W2, far from being eroded by W1 as materialists would have it, may actually be enhanced by it when the latter incorporates W3 objects that embody the fruits of technical change. It would follow, however, that the greater the constraints actually imposed by material and technological circumstances, the more compelling becomes the materialist position. The primitive societies discussed by Marvin Harris in his book on cultural materialism14 thus provide better exemplars for his thesis than post-industrial societies like our own. They are indeed more constrained by material circumstances than our own in what they can articulate and communicate, but no more than ours are their styles of representation actually determined by such circumstances. In this respect, a connectionist model of social interaction offers the degree of plasticity required by the emergent properties of collective representation that we are arguing for.
Design as a communication process
All artefacts that embed W3 products in W1 extend our communicative choices and freedom even if few of them are expressly designed to serve a communicative function. Some have communication purely as a by-product, while others – perhaps the majority – merge the communicative function with other attributes. Created physical extension systems, or artefacts that embody a communicative element, exhibit some degree of intentionality. Good design, for example, is nothing other than the ability, firstly, consciously to come to terms with and master this intentional element while respecting whatever constraints circumstances impose, and secondly to let the object or system of objects communicate that which we wish it to, that which it can, and that which it should – and no more. Yet it is a characteristic of all artefacts that over time they become established physical facts independent of their creators and hence take on a life of their own. To slightly misuse a Marxist term, they become ‘alienated’ and eventually migrate from the site of their creation.15 Artefacts encapsulate knowledge and store it either within their own physical forms or within the design traditions associated with their evolution – Abernathy and Utterback call the products of such traditions ‘dominant designs’,16 and when collections of related artefacts are involved, Freeman and Perez refer to them as technical systems.17 Some of these traditions can be highly coded and therefore easily transmissible; others, less coded or structured, can only be acquired slowly and in much more restrictive circumstances.18
Culture is a prime vehicle for the transmission of artefacts. Some of the knowledge they encapsulate, being explicit, can be transmitted in writing or through other artificial channels in a fairly impersonal way. Some of it, however, will remain implicit and can only be properly imparted over a long period of time and in face-to-face situations that we associate with the process of socialization. If impersonal transmission can occur through artificial channels and reach a large number of people in a short period of time, it nevertheless allows little scope for direct feedback and, like any form of broadcasting, therefore becomes a somewhat ‘hit and miss’ affair. Personal transmission, by contrast, uses sensory channels in parallel and for this reason can only reach a few people at a time. However, feedback here becomes possible and communicative intentions can be made clear by successive adjustments of codes and messages. The first approach thus offers greater coverage; the second greater control.
Three levels of communication problem
Cultural evolution, like biological evolution, is a matter of generating variety, selecting from it, and transmitting what has been selected to contemporaries and descendants. But its chosen vehicle for the generation and transmission of variety is artefacts instead of genes, information-bearing W3 products that have to survive the selective rigours of W1. How effectively these W3 products communicate, therefore – i.e., pass on their information content – is crucial to their chances of being selected and adopted, and hence to their future prospects for survival. What are the requirements of effective communication? Shannon and Weaver, in their classic text on the subject,19 identify three types of problem operating in any communication system:
| • | Level A problems: | How accurately can a given message be transmitted? (The technical problem.) |
| • | Level B problems: | How precisely does the message convey the desired meaning? (The semantic problem.) |
| • | Level C problems: | How effectively does the received meaning affect the conduct in the desired way? (The effectiveness problem.) |
Note that a problem at level A automatically implies a problem at levels B and C but that the converse is not true. And at all three levels, communication problems are greatly reduced by the spatiotemporal proximity of communicating parties. Whatever the level, the further one moves away from the here-and-now, where feedback and multichannel adjustments and corrections are possible, the more problematic communication becomes. Beyond a certain spatiotemporal distance one is led to switch from sensory to artificial channels. To those living in industrialized societies the need to switch may seem fairly obvious; yet it has not always appeared so compelling. Oral traditions still prevailing in certain parts of the world, for example, can handle intergenerational transmission processes tolerably well even though the anthropological evidence points to much information being lost en route.20 Oral traditions nevertheless suffer from spatial as well as temporal limitations that confine them to the transmission of the most central messages of a culture's repertoire. While they transmit messages exclusively through sensory channels, the high degree of selectivity applied to the message itself, together with the highly ritualized context in which transmission often takes place, makes oral traditions in practice a highly coded business quite removed from the more natural and sensory forms of interpersonal communication normally associated with spatiotemporal proximity.
Communicative efficiency versus communicative effectiveness
The switch from sensory to artificial channels will often be costly. New codes have to be learnt by all parties to a communication process and these may not be readily mastered. Furthermore, a communication infrastructure – transmitters, channels, receivers, etc. – may also have to be set up, thus further consuming scarce resources. The greater the resources that have to be devoted to the creation of a communication infrastructure, the more sensible it becomes to develop specialized codes that economize on its use. Thus, coding skills consist essentially in choosing that level of redundancy which minimizes transmission costs without sacrificing the clarity of a message. Conceivably, similar coding skills will be required for the processing of data within an individual E-space and for exchanges of data between E-spaces. If so, the connectionist perspective could be extended beyond the strictly neural level to cover the case of individual E-spaces acting as communication nodes within a social network. Relaxation and single shot algorithms would then become social information processing strategies that reflected the degree of cognitive cohesion present within a social system.21 At each level a balance would have to be struck between communicative efficiency – minimizing transmission costs without incurring a problem at level A – and communicative effectiveness, which might call for considerable volumes of technically redundant information for the message to be not only understood (the level B problem), but also acted upon, thus avoiding a problem at level C.
How can the balance between efficiency and effectiveness be operationalized? Shannon defines a message's relative information as H/Hm where H measures a message's actual information content and Hm the maximum amount of information it could carry.22 The ratio varies between 0 and 1 and represents the message's relative originality. By implication, the complementary magnitude 1 – H/Hm gives us the message's redundancy. Thus for a given channel we can assess the efficiency of a chosen repertoire of symbols – i.e., type case, vocabulary, and sentences in the case of a written natural language – for the transmission of a certain type of information.
Our coding scale (Figure 2.2) offers increasingly fast transmission rates and achieves ever-higher levels of compression of data into codes as one moves up along it; but at each point on the scale an efficiency-effectiveness trade-off will have to be faced. Taken in the purely technical sense (Shannon and Weaver's level A), redundancy will be a function of the statistical laws governing the collection of symbols and their relationship. Thus, for example, Shannon has estimated the redundancy of the English language to be about 50 per cent,23 and Moles, using a similar approach, puts the redundancy of the French language at about 45 per cent.24 In both languages, therefore, only half of what is said can be freely chosen, the other half being governed by the structural properties of the language itself – its syntax.
At the semantic level (Shannon and Weaver's level B), choosing an appropriate code is initially a question of establishing some measure of ‘fit’, or a mapping, between the statistical character of a chosen set of symbols and their interrelationships, on the one hand, and the statistical profile of the phenomena to be apprehended through the code on the other. Yet even if this is well done, it is still only half the story; for the chosen code, if it is to serve the needs of effective communication as well as of data processing, must also be known to the receiver. What repertoire of symbols is the receiver likely to have at his/her disposal? Would T.S. Eliot's recondite and coded references to classical mythology in his verse, for example, be accessible to the average contemporary reader? Did Eliot intend them to be?
Art and poetry pose the problem of effective communication in an interesting way. The artist's skill consists either in devising original high-level codes for describing and communicating phenomena, or in modifying existing ones in original ways. Yet inevitably, the closer artists get to achieving genuinely novel descriptions, the further they are likely to remove themselves from the coding conventions that allow them to reach their audience. Conversely, the greater the concessions they are willing to make to the coding needs and expectations of their audience, the greater the danger of vulgarization – that is, of resorting to a facile or commonplace coding of the phenomena they are trying to express. All works of art thus harbour an irreducible opposition between the artist's need for self-expression and the need to communicate.
The aesthetic experience, as we saw in Chapter 2, is one in which clarity is not always sought, the effectiveness of the message residing at least in part in its ambiguity. Powerful artistic statements are often those that have a certain tautness about them, keeping redundancy – i.e., rhetoric – to a minimum, and converting the resulting semantic difficulties into an enjoyable and instructive predicament for a given audience. Communicative effectiveness at level C, in such cases, is effectively achieved by not resolving all the semantic problems that might arise at level B.
Semantic versus aesthetic information transmission
Sensory channels offer an important advantage over artificial channels in that one can use them in combination with relative ease and for the most part almost unconsciously. This is perhaps just as well, since in our day-to-day encounters with our immediate environment, relatively few of the incoming messages that we register lend themselves to processing by a single sensory channel just for our convenience. Sensory stimuli are highly correlated in their sequencing and continually reinforce each other's internal probability structures: we first hear the train approaching the platform and then turn around and catch sight of it. If the train is close enough as it passes we may then feel the rush of air as it speeds past. Should the sound of the locomotive only reach us a few seconds after it had gone past we would be puzzled indeed.
Artificial channels are usually much more difficult to coordinate and combine than sensory channels. They are more cumbersome to use, and hence more costly in time and effort. For this reason any collapse of the multichannel experiences of the sensory mode into the single channel experience of the artificial one usually involves some sacrifice of data and therefore some compensating selectivity that preserves communicative efficiency in the way it is processed. In other words, the use of artificial channels usually calls for a greater degree of coding.
Switching messages between one sensory channel and another presents us with similar difficulties and also entails a loss of data. When can such data losses be countenanced? When they do not incur a concomitant loss of information; that is to say, when the underlying structure of what is being transmitted is sufficiently well understood or conceptualized that through careful coding or recoding it can survive the switch from one sensory channel to another – or for that matter from sensory to artificial channels. The abstract objects of W3, in so far as they express such a conceptual understanding, are thus better candidates than the concrete objects of W1 either for a transposition across sensory channels or for one from the sensory multichannel communication mode to the single channel artificial ones.
Messages that can be translated from one channel to another without undue information – as opposed to data – losses can be labelled, following Moles, semantic messages.25 They have a universal abstract quality that is accessible to anyone who can grasp their underlying principles. They tend to be more highly coded and articulate than what Moles calls aesthetic messages; these are untranslatable from one channel to another without information losses since what they describe are the concrete if sometimes ambiguous internal states specific to individuals. Aesthetic messages are thus confined to the channel or channels that transmit them and they are profoundly altered by any transfer from one channel to another. They have, as Moles puts it, ‘only equivalents, not equals’.26 Moles's views on aesthetic information are clearly convergent with those of Polanyi on personal or tacit knowledge.
In sum, we might say that semantic messages survive moves up the coding scale – particularly those moves that associate a change of code with a change of channel – by shifting away from the sensory channels that give us access to the concrete world on the left of the E-space, and, through successive acts of abstraction, by moving towards the more artificial channels that allow such a transposition of codes. Much data is shed on the way with consequent efficiency gains. Communicative effectiveness, however, is only maintained to the extent that information-preserving structures are successfully abstracted from such data before it is shed.
Aesthetic messages, by contrast, are required to preserve data to the extent that such data is required to convey the full richness and complexity of a concrete personal experience. Moves up the coding scale are possible and often necessary in order to give a minimum amount of outline and definition to a personal experience. Yet these moves are limited by the constraints identified by Moles: no transposition of codes is possible across sensory channels, and, to the extent that a personal experience is conveyed exclusively by means of such channels, aesthetic messages will tend to cluster in the south-west corner of the E-space. The location of each type of message in the E-space is shown in Figure 3.1.
Figure 3.1 Semantic and aesthetic messages in the E-space
Interesting consequences follow from the foregoing analysis which will be further explored in subsequent chapters. What we need to note here is that the more coded and abstract the epistemological objects that populate W3 can be made, the more adapted they will be to transmission by means of artificial channels. Thus not only, by dint of their abstraction, can they become independent of time and space in what they refer to, but in addition, by moving through artificial channels, they achieve a mobility and a diffusibility that further frees them from spatiotemporal constraints in their choice of audience. The concrete qualitative data that inhabits the south-west corner of the E-space does not enjoy such mobility and is thus more parochial in nature. Where it is communicable at all it is thus more confined, both in its choice of references and in its choice of audience.
In conclusion we might say that knowledge assets located in the north-east region of the E-space have the properties of ubiquity and universality while those located in the south-west corner enjoy those of uniqueness and identity. The closer one comes to securing one of these two sets of properties, the more of the other set must be given up. How does this affect the way that these different kinds of knowledge get shared? We turn to this next.
3.3: COMMUNICATING MEANINGFULLY: THE SHARING OF CONTEXTS
Technical and physical constraints on information diffusion
Figure 3.2 describes the diffusion of a given item of information in a target population of data processors as a function of time. The population chosen for the exercise could be of any size ranging from a handful of people to a nation-state or larger. It need not even refer to people necessarily: any entities capable of receiving, processing, storing, and transmitting information -dolphins, nerve cells, viruses, microprocessors, harmonic oscillators – could make up a diffusion population for our purposes. The target population is placed on the horizontal scale of the diagram and is expressed as a percentage; thus, for instance, on the left-hand side of the scale, only a small percentage is reached by a given message, whereas on the right the majority or even all of it is. Time is shown on the vertical scale. Clearly, the flatter the gradient of any curve, the faster the diffusion process at any given level of coding, and the steeper the gradient, the longer the time required for an item of information to diffuse through the population. If one assumes a population to be evenly distributed in physical space and homogeneous in its characteristics, then we might even expect the diffusion curves to be relatively smooth and continuous whatever their gradient. If, on the other hand, the population is structured into subgroups that differ in their spatial distribution and in their internal characteristics, then the diffusion curve may show a number of breakpoints as in Figure 3.3. In this second case, the first phase of a diffusion process, for example, might take place face to face in a primary group and the second phase through more impersonal specialized media. Such a two-step process might describe how news of a research breakthrough comes out of a research laboratory, being firstly communicated to work colleagues and then, after their comments and suggestions have been received, to a larger professional audience through professional journals.27
Figure 3.2 Diffusion over time
Figure 3.3 A non-homogeneous diffusion population
Available communication technologies also affect the slope of the diffusion curve. As might be expected, the larger and more spatially scattered the population to be reached by a given message, the more costly effective communication becomes in the natural mode and the more it will depend on the ready availability of artificial channels and an appropriate communications infrastructure.
Yet such an infrastructure imposes its own constraints on would-be communicators: limitations on the size and number of channels and on the availability of time slots for message transmission; coding restrictions designed to make efficient use of a given channel; queuing rules for various categories of message, etc. The social and physical configurations of a target audience – its internal structure and spatial distribution – thus interact with technical possibilities to establish the scope that exists for different types of communication. Where message transmission is deemed to be costly, a priority ranking system will be established so that certain audiences will gain privileged access to messages before others.
The transmission of meaning
The differential distribution of a society's stock of knowledge, whether or not it results from communication constraints, inevitably skews the initial endowment and subsequent evolution of individual E-spaces. The cognitive asymmetries that result further stimulate the process of differentiation, as well as a pattern of interaction between individuals and larger groupings that comes under the heading of the sociology of knowledge.28
Referring back to Shannon and Weaver's three levels of communication problems – the technical, the semantic, and the pragmatic – the sociology of knowledge to date has shown relatively little interest in the technical level, somewhat more interest in the semantic level, and the greatest interest of all in the pragmatic level, the one at which the problem of socially validated meaning appears. Recall that the technical problem as discussed by the authors is rooted almost exclusively in the physical characteristics of the communicative situation, and for that reason it is addressed largely independently of what is going on in the minds of senders and receivers. To be sure, these often adjust to physical communicative constraints by selecting suitable codes common to both. But in doing so they are then addressing semantic problems at Shannon and Weaver's second level, not their first. Meaning, by contrast, lies at the heart of Shannon and Weaver's pragmatic problem at level C. It is a wholly relational concept that unites sender and receiver; it is defined by Donald MacKay as
the selective function on the range of the recipient's state of conditional readiness for goal directed activity; so the meaning of the message to you is its selective function on the range of your states of conditional readiness.29
Otherwise stated, a meaningful message in some way changes an individual's disposition to act. Like coding, it affects an individual's probability structures but at a higher cognitive level. A communication becomes meaningful when it modifies the expectations that shape behaviour.30 It activates a receiver's cognitive and affective structures in such a way as to bring them into temporary alignment with those of a sender. Our sense of what is real is partly built up through multiple resonances of this kind, through the gradual acquisition, from infancy to adulthood, of meanings shared with individuals and groups that mediate our access to the life-world.31 To quote Sahlins:
the creation of meaning is the distinguishing and constituting quality of men – the human ‘essence’ of an older discourse – such that by processes of differential valuation and signification, relations among men, as well as between themselves and nature, are organised.32
In sections 3.7 and 3.10 we shall further develop the proposition that the level of abstraction and coding at which meanings are shared is consequential for a sociology of knowledge. As we have already suggested, translating experience into symbolic form does more than extend what can be talked about by allowing for remote reference: it also broadens a message's potential audience. Experience mediated through language and other coded forms is released from immediacy since signs become detachable from their context; they can be used outside the here-and-now to describe things that are absent to people who are not there.
Technology and the sociology of knowledge
Suitable coding gives access not only to new thoughts but to new technologies for communicating them. Technological and cultural evolution go hand in hand. If before Gutenberg's experiment with movable type in the mid-fifteenth century the transmission of knowledge required a strong spatial proximity between social groupings – indeed, Braudel has shown that even after the development of printing, the diffusion of books was confined to the channels through which the trade in luxuries flowed: the fairs of Lyon and Frankfurt in the sixteenth century, that of Leipzig in the seventeenth33 -today, with modern communications technology and for large tracts of human experience, the transmission process has become virtually instantaneous whatever the physical distance.
In principle, it should be new knowledge that flows most forcefully through the social system; such knowledge is likely to make the biggest difference to the prior expectations of receivers and hence carries the most information. New knowledge, however, often experiences difficulty in passing itself off as meaningful. Modern communications technology may overcome level A technical problems, and universal education may give us all a sufficient number of codes in common to skirt around level B semantic problems, but pragmatic level C problems can only be overcome by a mutual sharing of contexts, something that effectively gets more difficult with increases in spatial distance. A lack of shared context between sender and receiver thus becomes one of the main obstacles to the effective diffusion of meaningful innovation.34
3.4: SCANNING AND DIFFUSION AS SOCIAL PROCESSES
Invention and imitation
An early and sophisticated version of the thesis that cultural evolution was above all a communication phenomenon was developed by the nineteenth-century French sociologist Gabriel Tarde. In his book Les Lois de l'Imitation,35 Tarde argued that imitation was the most elemental communication activity and that it was to social processes what ‘undulations’ (i.e., vibrations or resonance) were to physical processes. With social evolution, Tarde claimed, imitation has become independent of hereditary transmission and hence of space-time constraints. Cultural diffusion did not necessarily require spatial or temporal contiguity, although clearly the denser the communication network the greater the scope for imitation and hence for cultural integration. A society, in this view, is nothing other than a group linked by a high degree of imitation or what he termed counter-imitation, i.e., a conscious collective striving to avoid imitating an exemplar which thereby confirms its identity in the repertoire of cultural symbols.36
But what, it might be asked, initially gets the imitative process under way? How does an exemplar worthy of imitation get to emerge in the first place? And how does it then get ‘selected’ for subsequent diffusion? Tarde's reply, reflecting his century, is essentially Darwinian. All repetitive imitation originates in an individual invention that survives a process of competitive selection, criticism, modification, and further invention.37
Invention, then, forms a bridge between individual and social processes. Although in our daily activities we imitate far more than we invent, invention can be thought of as the coding of a novel thought or experience which, under certain circumstances, will be subject to diffusion. One basic requirement that was discussed in section 3.2 is that the new creation be sufficiently coded to be communicable even if not always in verbal or written form. Forces are at work, however, which tend to blur the line that divides the initial act of invention from acts of imitation that subsequently diffuse it. When communication is ambiguous or otherwise problematic, where there is noise in the channel, when senders and receivers do not fully share the same codes or the same context, then information is lost and imitation is perforce incomplete. Consciously, or otherwise, the imitator then has to make up the shortfall with inventions and adaptations of his own.38
All imitation thus carries the seeds of new inventions. For this reason, following Schumpeter, we shall use the term ‘innovation’ rather than ‘imitation’ as it does not preclude that incremental inventive activity which follows diffusion and helps to adapt an original invention to the specific circumstances of an adopter.39 The steady accumulation of adaptive incremental inventions can often outweigh the novelty of the original invention to which they were responding.40
An invention is an act of encoding new knowledge that is performed in a single mind. Such knowledge remains undiffused on the left-hand side of Figure 3.2 until it is communicated; that is to say, it remains embedded in an individual E-space until externalized whether by accident or design. It becomes an innovation to the extent that it travels towards the right along the diffusion scale of Figure 3.2. The region of an individual E-space in which externalization occurs will establish an invention's initial communicability and hence its scope for diffusion, its clarity and force as an exemplar, and to some extent the need for further invention and adaptation. Once an invention has been externalized, it will spread through a population in a manner and at a pace that reflect the nature and extent of the coding process as well as the spatial and social characteristics of the target population. The social trajectory of a new piece of information or knowledge, whether embodied in objects, in documents, or in minds, is thus given by an interplay of internal and external forces that can profoundly modify a message's form and content in the course of its travels through a given population. Yet not only will the first and the last recipients of a given message in a population then gain access to different information, they will also bring to its interpretation quite different contexts and orientation.
Private and public coding
On the far left of Figure 3.2, coding activity, inventive or otherwise, is likely to be a mostly private affair. Indeed, it may not even be conscious. An individual inventor is rarely fully aware of the tortuous path that she followed in arriving at a new concept or idea. Much of it will be the product of a mental activity under no conscious direction.41 More importantly, the stored knowledge that the individual draws upon in articulating her experience on the left of the diffusion scale, and the way she combines various elements into patterns, appears to be largely idiosyncratic. She may be drawing upon the same external stimuli as her neighbours but each will make something quite different of them. The differences, for the most part, will be trivial and inconsequential. On occasions, however, they may become a source of radical new insight.42
As one moves towards the right along the diffusion scale, however, the knowledge and experience contained in individual E-spaces gets shared with an ever-larger percentage of the population and gives rise to collective forms of representation.43 It now becomes easier to communicate what one sees, hears, and feels since more of the knowledge that makes up the context of a communication has become common property. Coding performed towards the right of the scale will consequently appear less idiosyncratic and more consensual. While it may feel quite free of any sense of social coercion or pressure, it will nevertheless be channelled by perceptual and conceptual categories collectively held; what might then appear to an individual to be purely personal coding preferences will in fact be guided by values that he/she holds in common with different groups to which he/she belongs.44
In sum, towards the right of Figure 3.2 social constraints impose themselves on the act of giving form to personal knowledge. On the far left of the diffusion scale one sometimes comes upon the social deviant construing a strictly private world according to his own lights; on the far right by contrast one meets the conformist, largely dependent on others for the creation of his Umwelt as well as for some direction on how he should move through it.45 In the short term, society has fewer problems dealing with conformists than dealing with deviants since the former reinforce rather than threaten the established order. In the long term, however, a society without deviants deprives itself of the cognitive variety essential to social evolution.46
To call the horizontal scale of Figure 3.2 a diffusion scale is, in one sense, misleading and needs elaboration. In the next chapter, for example, we shall see that information flows from right to left as well as from left to right. At first sight this may appear puzzling since it seems to imply that knowledge which has already been diffused and is thus held by a large proportion of a given population can become suddenly undiffused again and unknown by the same population. Are we trying to run a film backwards?
Things become clearer if we bear in mind that what is flowing physically between data processors is not information as such but data and that each individual extracts information idiosyncratically from this data to construct her own schemes within her E-space. On the right such schemes are widely shared and on the left they are not, even though in both cases the same data may be used for their construction. The leftward movement in the diagram in effect describes a process of cognitive individuation through which an individual's perceptions or concepts gradually come to lose features that they had in common with those of her neighbours. Sometimes physical isolation may be the cause of such individuation; at other times it will be personality factors that lead an individual to build up and use her E-space in idiosyncratic ways. Either way, part of her cognitive resources shift leftward along the diffusion scale of Figure 3.2.
From coding to codification
Only the unique individual occupies the extreme left-hand point on our diffusion scale, and even then only episodically. The point is important enough to warrant a change in terminology. In this chapter and those that follow, therefore, the term codification will henceforth be used in preference to coding to reflect the fact that giving form to experience is rarely a completely autonomous process carried out by unique individuals located on the extreme left of the diffusion scale. It is at least in part a socially structured process. In effect, coding can be thought of as an idiosyncratic and limiting case of codification located at the origin of our diffusion scale. Suitable candidates for codification are those areas of experience – customs, laws, knowledge – that are stable enough to be shared with others and that express either a common group preference or the workings of power relationships within a group.47 As we have already seen, most of the information through which we initially build up our world is socially inherited and thus already codified.48 Berger and Luckmann put it thus:
I live in the commonsense world of everyday life equipped with special bodies of knowledge. What is more, I know that others share at least part of this knowledge and they know that I know this. My interaction with others in everyday life is, therefore, constantly affected by our common participation in the available social stock of knowledge.49
Yet this ‘common participation in the available social stock of knowledge’ remains problematic. For a start, only a small part of what passes for human experience is actually retained in consciousness, and whatever is held remains for the most part implicit and inaccessible. It can only contribute to the social stock of knowledge when it is in some way objectivated in a sign system, i.e., when it is codified to some degree. Then, there is the added complication that not all sign systems offer the same communicative scope. While anyone with a good command of a spoken language has a reasonable chance of being understood at least by his fellow countrymen, this is not necessarily the case with other forms of communication for which the required codes – whether these be abstract symbols, dress codes, or behavioural signs – may be the property of restricted groups.
Constraints on codification become simultaneously constraints on new ways of knowing and on sharing them. What is not objectivated in some kind of a sign system is exclusively available to the individual as personal knowledge within his own E-space. He can know his own experiences better than his neighbour can, having access to them in a way that the latter can never have. Incommunicable personal experiences of this kind are extreme examples of what Hayek has termed ‘local’ knowledge; that is, knowledge intimately tied to circumstances of time and place.50 Our myriad encounters with the world – a favourite walk, a fishing trip, a casual conversation, etc. – are of this kind. Even in our dealings with those who are co-present at an event, the element of shared experience offers but limited overlaps. Much stays largely local.
Personal versus impersonal exchange
What Alfred Shutz has called the life-world, our common intersubjective yet unspoken stock of shared experiences, is built up and continuously reaffirmed using information filtered through the codification constraints just described.51 On the most intimate level, usually the one we enjoy with a primary group, relationships have a spatiotemporal immediacy that is absent from our dealings with contemporaries. In face-to-face interactions, multiple channels are available which, together with the possibilities of giving and receiving feedback, and hence of correcting for distortions and information losses, preserve the richness of exchanges. They do so, however, only by keeping them local.
Contemporaries by contrast are not bodily present. In our dealings with them the here-and-now recedes into a bland anonymity as concrete experience is thinned out to render it communicable over larger regions of space and time. Such anonymity, of course, can also be injected into face-to-face relationships: the intimacy available in the here-and-now, after all, need not be much drawn upon. Market transactions, for example, have often been held to depersonalize face-to-face relationships, abstracting from the complexities of concrete individuals and situations, and codifying exchange into a featureless encounter between a buyer and a seller – in effect, transforming them into more manageable W3 objects.
It is in situations where the pressures of the social situation push for transactional efficiency that one can no longer react to individuals in their fullness; one is led to codify them into impersonal abstract roles that require a less discriminating response so that one can then react more narrowly to types: not to M. Jefferson who lives in the brick house in Chipstead Street, but to the baker, the ticket inspector, the newsagent, and so on.
Yet if anonymity can be injected in this way into a face-to-face relationship, an impersonal relationship can in turn sometimes be personalized. A rock star's fan club, for instance, is a shrewd commercial device for personalizing at least one of the parties in what can only be for the majority an anonymous relationship.
One can stratify communicative relationships, then, according to the degree of anonymity involved. The more anonymous the transactions between people the more ‘objective’ appears to be the meaning system that underpins it. At the highest level of anonymity, the world appears to be objectively given and self-evident, even to outside observers of the transaction. Anonymous relationships, the joint products of spatiotemporal distance and large transacting numbers, are more likely to be found on the right of our diffusion scale than on the left. Here too, it will be shared symbols – conventions that refer to abstract codifications – rather than shared experiences that contribute to the building of mutually intelligible worlds or components thereof. Shared experiences, particularly those that are the fruits of socialization within the primary group, have an important part to play in this construction but, being strictly local, they cannot involve contemporaries to any great extent. The reality that we can take for granted in our dealings with others – the life-world – will therefore of necessity reflect the numbers involved and by implication the degree of anonymity of interactions with them.
To summarize, each of us possesses knowledge whose characteristics would locate it at different points along the diffusion scale of Figure 3.2. Towards the left, as we have seen, the knowledge we possess tends to be more idiosyncratic, personal, and possibly proprietary. Some of it will remain in our sole possession, locked inside our skull and inaccessible even to our most intimate acquaintances. We take such knowledge with us to our grave and it perishes with us. Towards the right, however, such knowledge becomes an increasingly common possession: public, freely available, and, if codified, largely impersonal. Acts of communication can then move this knowledge either to the left or to the right along the scale, its mobility being a function of its codifiability. Knowledge that is hard to codify will move more slowly and in smaller volumes than knowledge that is easy to codify; the latter will always travel best along the scale.
Yet such mobility, as Shannon has taught us, is a purely technical property of well-codified knowledge. Barriers to movement along the scale exist at the semantic and at the pragmatic level.52 For the effective transmission of codified knowledge also presupposes the prior existence of shared substrata of uncodified meaning that can unite sender and receiver by placing them on the same ‘wavelength’. How far they can be united in this way is partly a matter of the spatiotemporal distance that separates them and partly of whether they are in fact intimates or merely contemporaries. Both Tönnies and Durkheim saw in the ability to communicate with contemporaries rather than intimates one of the driving forces of social evolution, shifting social and economic exchange away from the local and parochial and towards the universal.53
The authority of the transmitter
In the early phases of an individual's development, any idiosyncratic knowledge that he/she possesses on the left of the diffusion scale is likely to be overwhelmingly concrete rather than abstract.54 Yet, even though such knowledge remains highly personal and by implication hard to share, it will still be largely bounded by the abstract categories which the individual inherits from his/her culture and which are themselves widely available within it.55 In actually making use of inherited abstract categories, the individual typically begins life on the right of the scale; he/she may then move towards the left, but only as specialized concepts and experiences are acquired. Any position occupied by an individual other than at the extreme points of the scale, however, makes him/her a potential communication link in a scanning or a diffusion chain, picking up knowledge from others, possibly processing it a bit, and then passing it on. When the individual passes it on to those located to the left of him/her, he/she participates in a scanning process; when it is to those located on the right, in a diffusion process.
Playing a link role confronts an individual with new requirements. If he/she is the recipient of well-codified information, the decision to adopt and transmit it is made easier by the fact that the data is visible and can therefore be assessed for plausibility and internal coherence. As with the Ten Commandments, the very clarity and consistency of a message can make its adoption compelling. If, on the other hand, an individual has to grapple with uncodified data, his course of action is much less clear. Since the meaning of uncodified data has to be teased out of it and may only emerge slowly, much of it, at least in the early stages of a transaction, may have to be taken on trust. Polanyi illustrates the need for trust in the transfer of knowledge between a master and apprentice:
To learn by example is to submit to authority. You follow your master because you trust his manner of doing things even when you cannot analyse and account in detail for its effectiveness. By watching the master and emulating his efforts in the presence of his example, the apprentice unconsciously picks up the roles of the art including those that are not explicitly known to the master himself. These hidden rules can be assimilated only by a person who surrenders himself to that extent uncritically to the imitation of another.56
In cases such as the one above – and it typifies a large number of social relationships ranging from a Zen master with his acolytes to a father proffering career advice to his children – the power of the receiver resides essentially in her ability to grant or withhold her trust, to ‘surrender’, as Polanyi puts it, to the higher power of the transmitter. While the feeling of trust exists, the power of the transmitter, often charismatic in nature where knowledge is personal and uncodified, will dominate the relationship and keep lines of communication open between the parties. Nevertheless, what the receiver will actually construct out of what the sender transmits will remain indeterminate where uncodified data is involved. In such cases the act of imitation can never be complete and some of the work of codification and abstraction will be left for the receiver alone to perform – the topic of the next section.
Coercive and non-coercive codifications
Scheler has argued that in any society, all human knowledge is given as prior to human experience.57 Order and meaning are imparted to the individual from outside. The acquisition of language is a conditio sine qua non for gaining access to them; language, by coercively patterning individual experience, gives it a facticity that is rarely challenged. To adapt a term that we used in Chapter 2, once language has been internalized, it's single shot algorithms all the way. Powerless to resist the force of linguistic articulation, the developing individual locates himself at the outset on the right of the diffusion scale of Figure 3.2, with respect to both concrete and abstract knowledge. Personal efforts at codification, being socially determined, are unlikely to move him leftward towards more personal and idiosyncratic forms of knowledge until much later in his development, if at all. By then, of course, the die has been cast.
The flaw in Sender's formulation, it seems to me, resides in its over-reliance on language as the coding paradigm. The less articulate the code, the greater the discretion one enjoys in deciding how it is to be applied and what it is to be applied to – and by implication, therefore, the more one enjoys some freedom of choice in interpreting its various applications. Moreover, if, in a simple, undifferentiated society, socially mediated stimuli can be presented to the individual as mutually reinforcing and unproblematic, as soon as a minimum level of social complexity is reached stimuli emerging from diverse social groupings will often appear contradictory and destabilizing. The codes through which these are then filtered will lack their former apodictic quality.
Tarde,58 like Scheler, had also noted the coercive nature of social codes and the constraints that they place on individual freedom and fantasy. Yet in certain circumstances they lost their totalitarian character. Tarde claimed that in imitation, there were two despots, custom and fashion, and that of the two, custom was the more tyrannical. A hierarchical society, he argued, will be dominated by custom, a democratic one by fashion. With development, fashion replaces custom, for fashion according to Tarde is driven by rational selection processes and over time rationality will win out.
The key difference between custom and fashion is that the first is transmitted unilaterally from ancestors to contemporaries, whereas the second is transmitted interactively between contemporaries. Both originate in a codification that occurs at a given moment on the left-hand side of our diffusion scale. But in the case of fashion, the power of adoption or rejection resides mainly with the recipient. Fashion operates through a competitive process that results in a much higher rate of innovation than that which is possible in the case of custom. Where custom reigns, on the other hand, messages originating on the left tend to be accepted unquestioningly by those on the right. There is no way of communicating with ancestors and hence no way of negotiating with them. One can take or leave what they say, but if they are endowed with authority by the social order, then the tendency will be to take it.
With contemporaries, the situation is different. Novel messages emerging on the left are accepted or rejected on their intrinsic merit following large-scale sampling, testing, and deliberation by potential adopters, with the outcome feeding back to would-be innovators and trend-setters – a scanning process that now sets up a flow of information along the diffusion scale from right to left. Feedback in turn stimulates additional efforts at adaptation on the part of potential innovators, efforts which act competitively to limit the lifespan of any single fashion cycle and thus to weaken its dominion over the social order as new alternatives loom into view.
Multichannel versus single-channel communication
Whether one is dealing with custom or with fashion, the more abstract and codified messages become, the more likely they are to travel beyond the environment in which they originated and the greater the power they have of establishing remote reference. Conversely, less codified, concrete messages are more temporally and spatially confined. In pre-literate societies, for example, such as the !Kung's, the bushmen of the Kalahari, only the immediate context is available to children to learn from, and the lack of any broader institutional effort at codification leads to an almost total absence of formal teaching.59 The power of remote reference is therefore denied to !Kung children owing to the absence of an adequate coding repertoire. Recall, however, that efficient coding itself presupposes a shared context within which meaningful messages may be exchanged, even if such a context operates at a more implicit level than the coded message itself. Spatio-temporal distance reduces the probability of sender and receiver sharing a common context and thus requires that the codes used do more of the work required in getting a message across. This carries a cost. The contribution of a shared context to the effectiveness of even well-codified exchanges is not always obvious to members of western literate societies. Edmund Leach suggests that
In our own western, literate, mechanically organised society, so much ‘true’ Aristotelian knowledge is built into the cultural system that we mostly take it all for granted that logic of this kind is an essential component of common sense. Yet in practice, we only exploit formal logical principles in the relatively rare instances in which we are seeking to convey exact information at a distance using a single channel of communication, as in writing a letter or a book, or speaking to someone over the telephone. When two people are in face-to-face communication, so that they can use several channels of sensory information simultaneously – touch, sight, hearing and so on – the logical ordering of individual messages is much less obvious. If you record unrehearsed conversation on tape, you will find that on playback, very little of it is immediately comprehensible; yet in context, all those present would have understood what was being said.60
I read Leach to say that multichannel communication reduces our need as well as our ability to resort to artificial codes. Multichannel communication is communication in the natural mode, it is the coordinated deployment of gesture, speech, tone, clothes, movement, in the service of messages whose complexity would normally overwhelm the single channel. Communicating at a distance denies us the rich coding resources of the natural mode. Beyond a few metres, facial gestures lose their crispness so that, for example, stage actors need to use make-up to articulate (i.e., to code) the physical features of the characters they portray; they also have to dramatize relevant behaviour patterns to get them across – film, which unlike the theatre allows close facial shots, calls for very different acting skills. A few metres’ distance more between transmitters and receivers and microphones has to be used, voices raised, and many modulations of tone and pitch forgone to maintain the requisite level of verbal comprehension. More coding, more data lost. Now stretch the communication distance a fraction further and visual contact itself is broken. To the adoring throngs below, the Pope is reduced to little more than a white speck on the balcony of St Peter's. Here, the visual component of the message has all but disappeared, and the microphone, if used, is then left to say it all. With increasing distance, therefore, the natural multichannel mode of communication gives way to the artificial single-channel mode in which the need for efficient coding of necessity leads to a repression of expressive subtleties, unless these, in turn, can successfully be codified through stylistic conventions. Yet even then, much data is lost.
Abstraction as an alternative to shared context
Chapter 6 will explore the hypothesis that as western culture has extended itself spatially across the globe – and partly in consequence of this it has experienced the need to develop single-channel communication technologies – it has developed a marked preference for codified and abstract over uncodified and concrete forms of exchange with telling consequences for the functioning of its institutions. Here, we simply make the point that in overcoming the spatiotemporal distance that separates sender from receiver, efficient coding often only solves the technical communication problem by creating new problems at the semantic and at the pragmatic level where the issue of shared context looms largest. Messages are then efficiently transmitted but are not necessarily understood, or, if understood, they do not lead to the desired behaviour. We discover, sometimes rather late, that communicative efficiency and communicative effectiveness make uncomfortable bedfellows. Efficiency is concerned with the resources deployed to achieve a given communication effect – the coding effort involved, the channel capacity used up, the required transmission time, etc. Effectiveness, on the other hand, is measured by the degree of ‘fit’ that obtains between the outcome of a given communication act as intended by a message source and its actual outcome. The first is concerned with economizing on means, the second with achieving ends. Effective messages have to be both plausible and useful in the sense that their recipient will consider them to be substantially true and relevant to his/her circumstances.
A sharing of context by senders and receivers is one way of enhancing the effectiveness of a message while saving on data transmission as well as coding effort and time. An alternative approach available for economizing on communication costs while maintaining effectiveness, however, is to use coding categories that succeed in retaining their validity across multiple and variable contexts. This moves message content towards a higher level of codification and greater abstraction.
Abstraction is a crucial requirement. Highly codified messages, whether abstract or not, travel fast and far, spanning cultures remote from each other both in space and time. Their effectiveness, however, is not guaranteed since, particularly in situations where they are transmitted anonymously, they cannot easily be authenticated. They may be deemed potentially relevant; but who, or what, stands behind their claim to truth?
From a transmitter's point of view, this effectiveness problem is compounded by the fact that as an audience grows either in space or over time, it can be subdivided into groups with sometimes radically different approaches to the validation of truth claims. The Kuba of the Kasai, for example, take historical truth to be whatever is accepted by the majority as worthy of belief – a specification not unknown in those western cultures whose judicial system admits of trial by jury. The Tobriand Islanders, on the other hand, take historical truth to be whatever the ancestors claim to be true.61 In each society, consequently, acceptable messages are likely to emerge from quite different sources, to go through different authentication procedures, and to require a different context.
It can also be safely anticipated that the Kuba will examine messages less critically and will therefore be more vulnerable to communicative manipulations, whether for public or private purposes, than will the more conservative Tobriand Islanders.
A more political view of how truth claims are handled takes message acceptability to be a function of power relationships between senders and receivers, the ability to make people believe being rooted in the ability to make them conform.62 The ‘might is right’ position has two possible interpretations. The first, which we might call Orwellian’, is that a coercive use of power has a direct impact on an individual's cognitive capacity and that the use of force or the threat to use it can lead people to actually believe as well as to affirm propositions contrary to the evidence of their senses.63 The second interpretation, which we might term ‘Machiavellian’, is that power can be used non-coercively to control the flow of data along the diffusion scale and thus to regulate the access of various social groups to the different cognitive categories that can be derived therefrom. In the second instance, the exercise of power over cognitive processes remains indirect.64
Will such power always be used cynically and in self-serving ways? Marxists and non-Marxists will disagree on how far communication filters placed at different points on the diffusion scale can genuinely contribute to the detection of errors and to the systematic falsifications of unwarranted truth claims on the one hand – the justification usually given by power holders for attempts to control the diffusion of information through censorship practices – and how far they invariably end up in the service of dominant groups and of the cognitive order that they seek to promote on the other. Many non-Marxists will be quite happy to go along with the view that the control of information flows, at least on occasions, should also be justified on efficiency grounds.65
Universalistic versus particularistic exchange
The dialectical tension that exists between social and individual thought reflects the terms on which they transmit to each other. Marx, perhaps its most articulate observer in the nineteenth century, believed that the dialectic has a physical basis, that material and social conditions establish both the scope and limits for human thought.66 He believed that mental activity is founded in physical activity – i.e., labour, broadly defined – as well as in the social relations brought about by this activity.67 He termed physical activity the substructure, and the articulate mental world it gave rise to, the superstructure. In Popperian terms, Marx held that only W1 was causal and that W2 and W3 – the latter one being the superstructure – were both epiphenomenal. Yet if, instead of treating physical and mental activities as distinct categories, we broaden Marx's concept of physical activity to include data processing of the kind that takes place in an E-space, we arrive at a view of W1 and W3 quite at odds with Marx's own, and somewhat closer to that of Berger and Luckmann when they say that
It is an ethnological commonplace that ways of becoming and being human are as numerous as man's cultures. Humanness is socioculturally variable. In other words, there is no human nature in the sense of biologically fixed substratum determining the variability of sociocultural formations.68
The above comment, however, confronts us with a dilemma: for how can the reality of W3 proclaimed by Popperians be reconciled with the cognitive relativism implicit in Luckmann and Berger's views of human nature? It is after all the putative reality of well-tested W3 objects that gives a measure of legitimacy to the products of abstract thought. It is the reason why, as spatiotemporal distance increases in social exchange, we feel entitled to blend in a measure of abstraction with our codification in order to ensure communicative effectiveness. If concrete situational contexts lack a sufficient overlap to purge meaning of its ambiguities as exchange develops, then an abstract context can be developed for the purpose instead. In sum, the larger and the more spatially scattered the population brought into a communication nexus, the more important it becomes that this should be universalistic in its orientation rather than particularistic.69
Such an orientation, however, need not imply a realist stance. An alternative to the view that W3 objects have to be true to partake in the discourse of society is one which takes them to be nothing more than useful and reliable tools that can be applied to W1 problems. This pragmatic approach to abstract knowledge goes by the name of operationalism.70 The utility of knowledge is explored further in the next section.
Abstraction and diffusion
If, as Marx believed, W1 material and social conditions directly shape human thought in W2, unmediated by an autonomous W3 – he was, in effect, denying thought any emergent properties – how do they effectively impinge upon consciousness in W2? In the previous chapter we argued that the process was at best indirect and that W3, whatever cognitive autonomy it is actually granted, will help to dig the channels through which the data of the external world will flow as it irrigates our thinking.
Such data, however, does not reach all of us at the same time or in the same volume. Sociologists of knowledge claim that strategically placed power holders seek to regulate its flow by operating filters that control its distribution among various groups. They implicitly assume that in the absence of such filters, it would flow freely and voluminously in all directions. Hence the internal structure of the data – i.e., its degree of codification and abstraction – remains unproblematic for the sociology of knowledge, its effective distribution in a given population being almost exclusively determined by external factors such as social organization and power relationships.
Our own analysis of information and communication processes leads us to challenge the adequacy of a purely sociological approach to data flows and to offer an alternative augmented by epistemological considerations. Power relationships are not thereby occulted. Far from it. But the influence they exert over how knowledge is actually allocated to target groups is attenuated by distributional problems: the perishable nature of the commodity; the consequent need to package it securely for travelling; the extension of its shelf-life both at the wholesale (broadcasting) and at the retail (one to one) end of the communication channel, and so on. In sum, a blend of technical, semantic, and pragmatic problems ensures that a given item of knowledge at one time and place is not necessarily the same product it might be at another.
We start by bringing into an orthogonal relationship the abstract-concrete dimension of the E-space and the diffusion dimension discussed in this chapter. For expository purposes, we dichotomize each dimension as shown in Figure 3.4 to produce a fourfold classification of knowledge that reflects its degree of generality along one dimension and the extent of its diffusion along the other.71 We then obtain:
Figure 3.4 A typology of knowledge (1 )
| Quadrant 1 : | Knowledge that is concrete and undiffused: what Hayek has termed ‘knowledge of particular time and place’ – i.e., local knowledge.72 |
| Quadrant 2: | Knowledge that is abstract and undiffused: such knowledge may be lawlike in character but is the property of an individual or a small group – i.e., esoteric knowledge. |
| Quadrant 3: | Knowledge that is abstract and diffused: Boyle's law, the melting temperature of ice, etc. – i.e., scientific knowledge. |
| Quadrant 4: | Knowledge that is concrete and diffused: gossip, rumours, news of current events – i.e., topical knowledge. With repetition topical knowledge gives rise to legends, myths, histories, etc. |
This classification does not distinguish between codified and uncodified knowledge, and therefore leaves out of consideration one of the major variables that affects the distribution of knowledge along the diffusion scale. The relationship between codification and diffusion will be examined in section 3.11. What we want to examine here in somewhat more detail than in the previous section is how the diffusibility of knowledge might be affected by its degree of abstraction.
Exploring the quadrants
Quadrant 1 holds local knowledge. An extreme of this is the isolated, immediately given, concrete private experience, the reverberations of a mental event forever locked inside a single mind. It cannot move in any direction out of quadrant 1 without shedding its information content along with its data, and this whether it moves towards abstraction in order to achieve greater generality, or towards communicability (through coding) and hence greater diffusibility.
Quadrant 1 experiences may be meaningful but they cannot easily be shared. The knowledge they yield may have some utility, but while it is confined to the south-west region of the diagram, it can only be a local and personal utility accruing to the individual in possession of the experience, not a social utility from which others may collectively benefit.
Any increase in the social utility of quadrant 1 experiences would have to come from an extension of their scope, from an increase in the range of their potential applications. Utility can be increased in this way by a move towards greater abstraction and generality; that is, a move up the vertical axis of Figure 3.4 and towards quadrant 2.
Although we have so far placed only one individual in quadrant 1 for illustrative purposes, it is quite possible to locate intimate small groups in it, groups whose collective encounter with reality is likely to be idiosyncratic and parochial in so far as they draw their epistemological resources solely from this region. At times, non-rigorous validation procedures may help to bestow a high degree of social legitimacy to knowledge and experience emanating from quadrant 1, but again only at the price of a limited diffusibility that does not pose too rigorous a challenge to the procedures actually used. The hallucinatory experiences of the Yaqui Indian Don Juan and his circle as described by Carlos Castañeda, for example, gave them prestige and charismatic power in their community but perhaps only because it was a small one.73 A more familiar example is provided by the work of the sociolinguist Basil Bernstein who showed how the limited and highly situational coding practices of working-class British children has the double effect of implicitly shoring up the limited world they construct with their peers, whilst at the same time cutting them off from a wider, more impersonal and abstract universe of discourse that might ultimately challenge it.74
Quadrant 2 yields knowledge that, by dint of its higher degree of abstraction and generality, would enjoy a wide social application were it not the monopolistic possession of some individual or group – it has potential utility. There exists a strong social demand for such knowledge whether it is actually diffusible or not; but because it meets the two key requirements of economic value – i.e., it is scarce as well as useful – it is subject to appropriation.75 The mechanism of appropriation will vary with the codifiability of such knowledge as well as with the relative power of senders and receivers in the communication nexus. But except perhaps for the case of small, isolated pre-literate groups, there is usually a greater demand for the social diffusion of knowledge residing in quadrant 2 than for that located in quadrant 1. The audience for which it promises to be relevant is larger and its own potential utility is also more manifest. For this reason a horizontal diffusion of knowledge out of quadrant 2 and into quadrant 3, by dint of its higher degree of abstraction, is likely to be much faster and more certain than one from quadrant 1 to quadrant 4, and this in spite of whatever attempts might be made by its initial possessors to prevent or retard the process.
Concrete knowledge originating in quadrant 1 is for the most part highly parochial. In his field observations of oral traditions, Vansina has observed that many pre-literate groups only narrate their own history and refuse to recite traditions belonging to neighbouring groups, particularly where these are protected by rights of ownership. Illustrating his point with another Kuba example, he describes an event that occurred once in the Ngongo tribal council in Natumba, when a speaker was reprimanded by his listeners for reciting traditions belonging to the neighbouring Ngeende and had to break off.76
We need not reach out as far as pre-literate societies for evidence of a preference for concrete local knowledge. Even in industrial societies, a predilection exists for news that is topical and focused on the immediate community rather than news which describes more general trends and developments that occur elsewhere. Abstract issues, to be sure, are of some interest, but only when they can be brought down to earth and their relevance for local concerns made manifest. A preoccupation with the environment in western industrial societies, for example, only became pressing when people could start relating what was being said in remote international forums on the subject to concrete realities such as polluted rivers and oil-smudged beaches that they could experience directly for themselves.77
It is not just the limited interest that concrete knowledge has to offer outside its immediate locality that limits its diffusion from quadrant 1 to quadrant 4. The information losses incurred in its transmission must also rank as an important factor. Recall from Chapter 2 that abstraction is one way of economizing on data to be transmitted, codification (alias coding) being the other. Although we defer a fuller treatment of how codification affects the sharing of information to section 3.11, here we note that to the extent that concrete local knowledge is less likely to be codified than abstract universal knowledge – this was one of the conclusions of our earlier discussion of the E-space – it will tend to diffuse more slowly, and will do so through a process of personal contact rather than through speedier and more efficient means.78 The transmission of many intangible skills, for example, requires intense personal multichannel contact between teacher and pupil. Polanyi maintains that given the right degree of contact, one can in fact become a proficient performer simply by following a set of rules that one may never become fully conscious of.79
The fragility of moves towards abstraction
Imitating a teacher is a slow business and often an uncertain investment requiring a great deal of practice and repetition. Yet without these, according to Tarde,80 science becomes impossible. For continuous practice and repetition furnish a statistical base capable of corroborating acts of abstraction or generalization. In the case of intangible skills such as piano playing or painting, the outcome may be nothing more than a well-tested rule of thumb or heuristic that cannot be further explained – in effect more a case of codification than one of abstraction as such; but with other activities, repetition and practice may yield intelligible principles which extend understanding to new areas.
A move towards greater abstraction thus offers the prospects of genuine gains in utility.81 We anoint this insight by labelling the diagram of Figure 3.4 a Utility space or U-space. The insight must be applied with care, however. For moves up the U-space may also offer no gain in utility at all: the statistical base accumulated through repetition and practice may be too flimsy to sustain an inference, or the inference itself may turn out to be faulty. The history of science abounds with examples of spurious generalizations that initially appeal but that do not stand up to the test of time. The failure of the pioneers of hypnosis from Mesmer to Braid comes to mind as one example of inferential inadequacy. Hypnosis only gained popularity when an adequate hypothesis – i.e., inference – was put forward to account for and bring order to the facts that had already accumulated.82
Yet if a move up the U-space towards abstraction and the creation of W3 objects is provisional, as Popper claims it is, and always subject to revision,83 does this not undermine its presumed utility? Is it then appropriate to allow the diffusion of potentially fallible abstract knowledge into quadrant 3 where by dint of its scope it might cause extensive damage?
In science, where the problem is most acutely felt, the institutionalization of validation procedures is designed to minimize any diffusion either within or beyond the scientific community of faulty abstract knowledge. Scientific validation procedures themselves, however, whether sponsored by the transmitter of knowledge or by its recipient, often express a particular philosophical commitment, whether this be to realism, to operationalism, to epistemological anarchism, or to whatever ‘ism’ happens to be topical. Despite the confusion, the most robust validation procedure will remain the performance of abstract knowledge in W1.
Rounding off our discussion on the U-space we observe that the propensity to abstract and the propensity to diffuse, interact. If abstract knowledge, by dint of its greater utility, finds a wider potential audience than concrete knowledge, the increasing size of such an audience and its active involvement with the material being diffused itself offer greater opportunities for practice, repetition, inferential learning, and validation. The two movements out of quadrant 1, towards abstraction and diffusion, thus feed upon each other. It should not be assumed, however, as we have already seen, that there is a generalized tendency for all types of knowledge to move towards quadrant 3. The move along the abstract dimension is often blocked by the coding propensities or capacities of a given group or culture. Diffusion, in turn, may be retarded or brought to a halt by discontinuities in the social fabric.
Knowledge thus gets partitioned into what Schutz called ‘finite provinces of meaning’.84 What accumulates in quadrant 1 constitutes a source of social differentiation; what ends up in quadrant 3, by contrast, acts as a source of social integration. Within a given group, however, integration is often only brought at the price of its own differentiation from a larger one. Only by sustaining a continuous movement of knowledge towards the north-east region of the U-space, together with a steady accumulation of universal abstract knowledge therein, could one possibly talk of the emergence of a single, widely shared world view. Societies, no less than groups or individuals, adopt various configurations in the U-space, partly in response to forces set in motion by their own internal development. These forces create vast tidal movements of knowledge both within and across their boundaries whose ebbs and flows are continuously remoulding the cognitive maps of their members – those at least not landlocked by tradition.
Does the foregoing commit us to the view that all epistemological progress is ultimately illusory and that human knowledge is irreducibly contingent?85 Or do the tidal movements of knowledge described above throw up geological formations over time that are more durable? If so, how durable? We discuss this next.
The language of participants versus the language of observers
Husserl maintained that the techniques of the natural sciences could not apply to social acts. Social acts involve meaning and meaning can only be grasped intersubjectively as ‘lived experiences’.86 Intentional acts, according to Husserl, cannot yield purely abstract and objective knowledge; and, given the need for context in the decipherment of meaning, nor can they be communicated to those who do not share a common base of experience. In other words, an inability to share meanings, whether due to a lack of common codes, common backgrounds, or common values, prevents or retards the flow of knowledge from quadrant 1 into quadrant 3.
Anthropology acknowledges the problem of meaning in the distinction it draws between emic and etic description (see section 3.2). With the first, a native informant is made the judge of descriptive adequacy: a given set of observations and conclusions will be considered valid if they are acceptable within his community. Etic description, on the other hand, places the power to arbitrate on descriptive adequacy with an external observer. A native narrative then becomes just another object of description. At issue is whether the descriptive terms employed by the external observer are those that other outsiders, placed in similar circumstances, would have used.
The problem of meaning, of grasping context, is believed to create an epistemological bias in the social sciences in favour of quadrant 1, of the concrete and the local, or in Windelband's terms, away from the nomothetic and towards the idiographic.87 At stake is the possibility of moving out of quadrant 1 without incurring unacceptable information losses. A shift into quadrants 2 or 4 decontextualizes knowledge and robs it of meaning. The belief that the natural sciences are free from this bias is due to a widespread assumption that unlike the social sciences, the natural sciences deal exclusively in universal codes and etic descriptions.
The cultural anthropologist Edward Hall has questioned the value of the emic/etic distinction by claiming that it is not possible to describe a culture solely from the inside or solely from the outside. As an alternative, he grades communication possibilities according to how far a common context can be shared between participants themselves on the one hand, and between observers and participants on the other. He cites the language used between airline pilots and control tower personnel as an example of what he calls highly contexted situational dialect. A shared knowledge of the operational situation by the actors allows them to use a language of great parsimony and low ambiguity in circumstances that are clearly defined and highly restricted.88
Like the East End children studied by the sociolinguist Basil Bernstein, airline pilots and control tower personnel communicate through a highly restricted code89 which economizes on data transmission when communication resources – channel capacity, attention span, etc. – are scarce. Their language might be considered emic in so far as an external observer might have to master both their code and the context in which it is used in order to make sense of what he hears. Much of their discourse, however, derives its operational effectiveness by drawing upon technical knowledge and concepts located in quadrant 3, so that a familiarity with local circumstances will need to be complemented by a more abstract and universal non-local type of knowledge, as well as by some mastery of the etic codes which give access to such knowledge.
In the Husserlian perspective, the purely external observer of social acts is a figment of the positivistic imagination. A minimum dialogue is necessary if an observer is to gain access to his subject and ‘a dialogue can be sustained only if both participants belonging to a community accept on the whole the same teaching and tradition for judging their own affirmations. A responsible encounter presupposes a common firmament of superior knowledge.’90 In our daily commerce with others, we unconsciously assume an ongoing fit between our respective meanings, between our world and theirs. In the natural attitude, we take our world to be unproblematic. Where a problem does arise between us, the natural attitude may have to give way to the more conscious theoretical attitude, but this will then require a special effort on both our parts. We may have to cease assuming a ‘common firmament of superior knowledge’ and begin to depersonalize certain aspects of our relationship. Each of us may then become aware of the other as an object of study as well as a co-participant in an exchange and may consequently hold back from an unrestricted ‘we-relation’. Pushed to extremes, however, such circumspection removes us from the communication nexus altogether and converts the newly objectified ‘other’ into a scientific specimen that yields only data rather than meaningful messages. The other, in effect, then becomes little more than a statistical input into a nomothetic generalization.
Competing codifications
A moment's reflection, however, will show that the need for a meaningful dialogue with the subject of an anthropological observation hardly suffices as an adequate basis for the distinction that Husserl wishes to draw between the social and the natural sciences. For does not the familiarization of, say, a young biologist with the abstract and etic codes that are the tools of his/her trade inevitably pass through a dialogue with peers, custodians of a particular scientific tradition that constitutes their‘ common firmament of superior knowledge’? Is the universality claimed by scientific discourse so uncontaminated by traces of its origins, of the concrete and the local circumstances that shaped its codes and its subsequent trajectory in myriad individual E-spaces, that we can declare it wholly free of emic description?91
If the relativization of knowledge calls into question the concept of scientific progress as objectification – and we do not prejudge the issue here – then such questioning surely applies to the whole of science and not just to the social sciences alone. As an ‘observer’ the nuclear physicist is as ‘social’ as the field anthropologist so that the ‘hard’ sciences no less than the ‘soft’ ones deal in categories that are socially mediated. In both, encounters between individuals who do not share the same context become potential sources of cognitive discrepancy and faulty communication. Discrepancies that occur in everyday matters can usually be accommodated within an existing, epistemologically undemanding, symbolic universe. We do not consider them cognitively problematic and what we call our commonsense knowledge is sufficiently pliable, with a little stretching and pulling, to absorb them without hiccups. Yet every symbolic universe is incipiently problematic. Sooner or later one crosses domains of experience where its writ ceases to run. It is in these lawless territories that rival and possibly quite deviant symbolic schemes may emerge, and if these are adopted, objectivated, and then diffused within the social fabric, they may come to develop a life of their own and one day rise to challenge the established institutional order itself. Since socially well-tested and objectivated knowledge is designed to yield compelling propositions about reality, any radical alternative to the existing symbolic order appears in its early formative phases as a departure from reality, a deviation, in extreme cases possibly a depravity or an illness.92 The social system will move to neutralize it either by downgrading its ontological status, by making a special effort to incorporate it in existing schemes, or by attacking it outright.
Where this happens, new knowledge may move up the U-space from quadrant 1 to quadrant 2 towards greater abstraction, but will then find itself blocked from diffusing into quadrant 3. It may have great potential utility and this for a large audience; yet if it is perceived as deviant this utility may never get exploited. Scientific methodologies do not do away with such blocking mechanisms. At best they make them less arbitrary by subjecting new abstract knowledge to more rigorous and rational testing procedures prior to diffusion than might otherwise be used. This has certainly helped to speed up the flow of new knowledge, although even in the most developed societies the filters that are erected between quadrants 2 and 3 are never wholly free of what Munz has termed ‘Social bonding’.93
Why does the emergence of a new abstract order constitute such a threat? And correlatively, why do existing practices, customs, habits, and perceptions display such inertia and resistance to change? Among the many possible ways of approaching these questions, a promising one seems to lie in examining more closely the nature of codification.
Every new act of codification involves both an affirmation and a negation. To allow alternative codifications of reality to coexist or to compete with each other is in effect to live under greater existential uncertainty or ambiguity and to accept a lower degree of ontological structure than many social systems can tolerate.94 In the neurological language of Chapter 1 they seek to avoid the conflict-laden explorations of relaxation strategies in their social transformation processing activities, preferring to apply single shot algorithms wherever they can. The distribution of power, authority, wealth, and status to their members requires stable, well-structured, and unambiguous cosmologies for their legitimation. It is to protect a brittle cosmological order that some societies impose such strict controls on new codifications. And even in systems where new ideas flow more freely, their diffusion will still remain a slow process of overcoming entrenched a priori presumptions in favour of an existing order and of bringing about a reconfiguration of prevailing beliefs and motivations.95
Trajectories in the U-space
According to Kelly, ‘social psychology must be a psychology of interpersonal understandings and not merely of common understandings’.96 Can common knowledge emerge from interpersonal understanding or does it require common understanding as a precondition?
A person's or a group's cognitive assets can be distributed across the U-space and assessed for how far they are shareable with others semantically and pragmatically.97 The south-west corner of the U-space speaks of a private, possibly idiosyncratic world that is in certain respects incommunicable; the north-east corner of an objective world that appears to be shareable with others unproblematically. The question of how objective and how far shareable turns upon the epistemological and social conditions which are imposed on any attempt to move out of the somewhat solipsistic universe of the south-west region. Common knowledge resides in the northeast corner of the U-space. Yet only if the individual trajectories that link it to the south-west corner are identical can we properly speak of common understanding.
A relativistic stance holds that any trajectory up the U-space towards greater abstraction is as good as any other since the criteria by which these moves are judged are socially determined and hence contingent.98 Furthermore, the abstract constructions available to us once we reach the northeast region of the space are also socially conditioned; they are possibly constrained by W3 but it does not follow that they are descriptive of it. Unlike materialists, relativists do not argue that only W1 is real and that W3 is all epiphenomena; their claim is that real as it may be, W3 can never be objectively known and that given the socially contingent nature of moves up the U-space, no progress towards knowing it better is possible.
Realists such as Popper, by contrast, take the position that while W3 can indeed never be known completely, progress towards knowing it better remains possible. A movement towards greater abstraction therefore, providing that it has been rigorously corroborated through testing, is not just a cognitive convenience that saves on data processing – essentially the operationalist view – but is also epistemologically meaningful.99 A trajectory through the U-space that leads upwards and to the right is then for realists a move towards a shared objective world.
The move does not, however, take place in a vacuum. Certain forms of social organizations and their underlying belief systems will speed up the movement towards the north-east region, while others will slow it down or block it altogether. In so far as the distribution of power in a society affects its choice of organizational forms and belief systems, it will play an important but not exclusive role in shaping the evolution of knowledge in individual U-spaces.100 We explore the issue of power in the next section.
3.8: THE SOCIAL DISTRIBUTION OF POWER
Codification as intellectual leadership
According to Tarde, to be social is to imitate.101 Pure invention, in contrast to imitation, requires a certain freedom from social pressures, a minimum ability to escape society. Recall, however, that an inventive codification is simultaneously an affirmation and a negation, an act of selection in which the state that is chosen or coded for gains an identity at the expense of the larger number of states that are consequently rejected. Where innovative coding involves a kind of social competition between alternative states, each being sponsored by a powerful individual or group, the successful diffusion of one will both stimulate and inhibit the production of rival states.102 The publication of a patent, for example, may place control of the diffusion of new technical knowledge in the hands of its creator and thus inhibit its unrestrained use; yet by demonstrating that a given approach is feasible, a patent effectively invites competitors to abandon alternative approaches and to devote more resources to the challenge of ‘inventing around’ the new patent. Thus by reducing technical uncertainty in one area, a patent may create a growing network of affiliations and cognitive commitments -’connection strengths’ in neurological parlance – based on demonstrated feasibility.103 Patenting, then, as an act of codification, pits the ability of originators to control the flow of new knowledge against the mimetic abilities of followers.
What is true of technology is also true of science. Yet in science the selective emphasis is perhaps placed more on abstraction than on codification as such, an emphasis that often leads us down a quite different diffusion path. A technology gains adherents through a concrete demonstration project covering a limited portion of reality within which it is shown to ‘work’; a scientific theory, by contrast, gains popularity if it can be shown to explain more than competing alternatives. Both technology and science achieve their effects in W1, but in the case of an abstract scientific theory, a W3 product, the test is usually at best indirect, inconclusive, and always subject to revision.
Not only is the testing of a scientific theory more problematic than the testing of a new technology, but the stakes are often much higher. For in science what are sometimes pitted against each other are not alternative physical contrivances, each with its own specific performance parameters, but competing and at times incompatible world views with ramifications that extend into the social order as a whole. The usurpation of a Ptolemaic cosmology by a Copernican one, for example, or of the creationist view of nature by Darwinian natural selection, in each case ushered in social and religious transformations that extended far beyond the specific and limited hypotheses in which they originated. Yet the lack of any scientific test that would be accepted as conclusive by opponents of the new ideas meant that decades had to pass – indeed in the case of the Copernican hypothesis, nearly two centuries – before the social and intellectual conflicts they engendered finally abated.104
Invention versus innovation
The ability to reduce cognitive uncertainty through codification in a way that is acceptable to others is a creative act of leadership and a source of power that sets those located on the left of our diffusion scale apart from those located on the right.105 The key term here is ‘acceptable to others’ since many individuals located on the left may be highly productive sources of new and original codifications which subsequently turn out to be too eccentric or deviant to diffuse beyond a small group of acolytes. These are often social marginals in the system, a creative humus from which may emerge the tender shoots of valuable new ideas.
One of the major achievements of the late nineteenth century and of the twentieth century as a whole has been to capture the creative potential of such people through organizational innovations such as the research and development laboratory which, in effect, decouple the process of generating new ideas from that of diffusing them or of exploiting them.106 Recast in evolutionary terms, the process of generating variety was institutionally separated from the process of selecting from it. For this to occur, invention had to be clearly distinguished from innovation.107 An invention describes the first instance of creative coding or abstraction whether it is to produce new scientific knowledge, a technical device, a new aesthetic trend, and so on. It invites but does not actually compel imitation. Innovation, on the other hand, covers the process of adopting an invention that has already taken place, the taking up and passing on of a new idea. The two terms interfuse, however, when the act of adoption requires an effort of adaptation that is itself inventive.108 This is more likely to happen when the respective loci and contexts of invention and innovation are far removed from each other.109
Science and technology are not the only sources of innovation and new knowledge. If, as the saying goes, imitation is the sincerest form of flattery, it is often just as likely to target an innovative personality as new ideas or contrivances. Imitation is a form of submission to an exemplar. If the object of imitation is a person rather than a thing, then the act of submission amounts to a recognition of personal authority and prestige rather than a pursuit of material utility, narrowly construed. One tries to emulate people one respects and admires.110
Who is recognized as an opinion leader in a social grouping will have an important bearing on what new codification or inventions are likely to be selected for imitation and on who is then likely to be doing the imitating. As de Tocqueville noted, in an aristocratic society, the object of emulation is one's hierarchical superior; in a democracy, it is likely to be one's neighbour.111 In both cases, the tendency will be to imitate those who are felt to be within reach. Where the authority and prestige of a group is threatened by uncontrolled imitation, however, it may be coercively restricted to members of the group alone. The Indian caste system provides examples of behaviours, clothes, and rituals which get appropriated by particular socioeconomic groups and the diffusion of which subsequently becomes strictly regulated.112 Social structure thus simultaneously influences the propensity to invent and to adopt and often prescribes the channels that link the one to the other.
The power of diffusers versus the power of adopters
The decision to adopt is sometimes a risky business, riskier perhaps at times than the act of creating itself which in most cases can be kept discreet if not actually covert.113 Acts of adoption performed by an opinion leader constitute selection judgements on which he takes his reputation and his prestige. He gambles what Bourdieu refers to as his symbolic capital in return for followers, measuring his success by their number and quality.114 Where the social order is conservative, that is to say where the prevailing cosmology promotes stability and tradition over evolution and change, the judgements of opinion leaders in matters of adoption will most likely be negative, as will be the perceived returns to innovation. In such cases, the established hierarchy of social prestige can block innovation and sometimes invention as well and may on occasions resort to coercion in order to do so.115
In sum, the diffusion or non-diffusion of knowledge is linked in important ways to the distribution of power in society. At times, established power relations actively favour innovative diffusion where this serves to reinforce the prestige and authority of a dominant group: the French language, for example, first gained its ascendancy through the influence of the French court in Paris; it then spread throughout the Ile de France during the Middle Ages, all competitors being relegated to the status of a. patois.116 In other circumstances, entrenched interests will be threatened by the spread of new ideas and will block them: the failure of Jesuit missionaries in sixteenth- and seventeenth-century China to successfully diffuse Christian doctrine was due in great measure to their inability to consolidate their power base at the Ming court in the teeth of opposition from the literati following the papal interdiction on the preaching of the new Copernican doctrine. Thus is the diffusion of certain types of knowledge promoted and the flow of competing alternatives restricted. All social systems have their information hoarders and their proselytizers; their respective effectiveness will depend on where they are placed in the system and on the institutional resources they command within it. It seems reasonable to assume, however, that power holders will seek to position themselves towards the left in the U-space, where not only will they be among the first to receive new knowledge when it emerges but also they will be well placed to control its subsequent diffusion to those placed to the right of them.
Gabriel Tarde was one of the first to link a social system's capacity for development to the power distribution within it.117 An early exponent of what later came to be known as the Convergence Hypothesis, Tarde argued that the moral and aesthetic community towards which the individual orients his behaviour was continually growing and that the dynamics of imitation were increasingly pushing for the adoption of common habits, customs, etc. He put this down to the spread of rationality and the growing power of fashion to overcome the conservative forces of tradition. Tarde's vision of societal development – for a vision it was – can be interpreted in our U-space as a gradual shift of society's cognitive assets from quadrant 1, where knowledge is parochial and its diffusion is either limited by local circumstances or subjected to strict controls, and into quadrant 3; there knowledge has become universal and widely available throughout the social system. Invention and imitation help to accelerate this shift of cognitive assets in the U-space, a process which has reached its most advanced stage, many social observers argue, in western industrialized societies.
We shall discuss the Convergence Hypothesis further in Chapter 6. We end this section by noting that any shift of knowledge in the U-space of the kind hypothesized by Tarde has been much assisted by the development of institutions explicitly designed to bring it about. What might be the contribution of such institutions to the diffusion of knowledge? The phenomenon of institutionalization is given a full treatment in Chapter 5. In the next section we confine our attention to the information processing role of institutions.
Economizing on social exchange through institutions
The relationship between knowledge and its social base is a dialectical one;118 that is, knowledge is at the same time a product of social change and a stimulus for it. But their interaction does not proceed smoothly. The move from one quadrant of the U-space to another is therefore a hesitant, fitful business with frequent pauses that may indicate the presence of zones of stability and inertia in the workings of the dialectic, zones in which novel abstractions either challenge the prevailing social order or are further consolidated and integrated into it.119 If a lack of movement in the U-space points to conflicts and blockages rather than consolidation and integration, it may be due to a confrontation between incompatible cosmologies that hinders the effective metabolizing of both existing and new knowledge. Movement or the lack of it may be given a number of interpretations that we explore in the next chapter. Here we note that where integration and consolidation rather than conflict occur in a given region of the U-space, we may expect to find institutionalization processes at work, building on the recurrent cognitive and communicative practices of individuals or groups.120 The codification and generalization of recurrent social processes both facilitate and are greatly facilitated by the existence of a stable institutional order. By economizing on information processing transmission and storage, institutions release social and cognitive energies that can be more profitably redirected towards non-programmed activities.
Institutionalization offers large-scale economies of mental and physical effort, but only by paying the price exacted by any act of codification or abstraction, namely, a suppression of competing formulations. Institutions are in effect W3 objects and share with other W3 objects a certain facticity that masks the epistemological fragility of their foundations. To survive at all, institutions need to build up a certain level of structural inertia which makes them resistant to change. They achieve their inertia by a subtle application of coercion to social behaviour. Effective socialization to the institutional order, however, often conceals the latter's essentially contingent nature, allowing coercive measures to be applied selectively and economically; even behaviour that appears to occur quite spontaneously will for the most part have been gently channelled through the appropriate institutional forms. Stability will prevail in those areas where the institutionally prescribed order can be taken for granted and no competing alternative is in sight that might rise up to challenge it. Social action is then at its most predictable and controllable. Social marginals, to be sure, those who by definition have the lowest stake in the established institutional order, may attempt to corrupt or disrupt it, but unless they either form a sizeable minority or can offer an appealing alternative to more centrally placed groups, their strivings will remain peripheral and fruitless.
Institutions can only be constructed from a set of shared assumptions or world views that are accessible to all the relevant actors. They do not emerge spontaneously within the body politic but from a gradual articulation of shared typification that is itself slowly built up by accretion. Institutions, then, represent a long-term collective investment designed to lower the information costs of social exchange – what institutional economists, more narrowly focused on economic processes, label transaction costs.121
From what has just been said, it will be apparent that in U-space terms effective institutions cannot really be quadrant 1 phenomena. They express universal aspirations and aim at a wide-ranging consensus. In effect, they may best be thought of as relay stations on any one of a number of imaginary paths that link quadrant 1 to quadrant 3. One might even hypothesize that the closer they are located to quadrant 1, the more institutions will express the exclusive perceptions, values, and interests of particular groups. Conversely, the greater their proximity to quadrant 3, the more they will reflect the broader concerns of society as a whole.
The distinction we are drawing here between particularistic and universalistic institutions can be traced back to Parsons.122 Societies with a universalistic orientation are less likely to suffer the pressures of social bonding that slow down or arrest the flow of knowledge from quadrant 1 to quadrant 3. The degree of abstraction associated with universalism imposes a degree of impersonal discipline on institutional processes that mitigates somewhat the arbitrary exercise of power associated with social bonding.
Controlling knowledge flows: the institutional legitimation of the social order
Power in society is, among other things, power to define and impose upon others the critical socialization processes upon which institutional integration can subsequently be built and by means of which a description of reality compatible with the power holder's own interests can evolve. Power holders aspire to institutional positions in the U-space that allow them to filter and regulate the information flows that reach different groups. Through control of the mechanisms of abstraction and diffusion they are in a position to determine what shall flow out of quadrant 1 and into quadrants 2 and 4, and thence ultimately into quadrant 3. The potential social utility of new knowledge will often only effectively be exploited therefore where it turns out to be aligned with the personal utility of key power holders where this is expressed in specific institutional practices.123
Control of knowledge flows in the U-space by power holders can be maintained as long as groups offering competing alternative and possibly quite deviant cognitive schemes cannot dislodge them. The threat of such destabilizing intrusions explains why, in most societies, access to the control of knowledge flows is institutionally regulated. In a pluralist society, regulation may be less restrictive than in a totalitarian one and access to the control mechanisms may be made subject to competitive processes. Nevertheless, the flow of large volumes of socially useful knowledge will remain subject to the monopolistic influence of groups strategically located in the U-space. What then remains available for processing by individual E-spaces is inevitably affected.
Is the control of knowledge flows necessary to the stability of the social system itself? Or merely to the survival of power holders within it? Unbridled rivalry between competing definitions of social reality produces a fragmented social order and transient constellations of subjective realities and identities. All worlds are then seen as relative, including one's own. The feeling that ‘anything goes’ sooner or later leads to a crisis in the perceived legitimacy of the established institutional order. As Berger and Luckmann put it, ‘all social order is precarious ... all societies are constructions in the face of chaos’.124
Legitimation of the social order aims to keep such chaos at bay. Competing claims, however, by weakening the collectivity's commitment to any single legitimate construction, bring chaos one step closer – unless, that is, the competitive process itself can be regulated with reference to some stable meta-order. A stable meta-order is what science aspires to provide in the case of secular western societies, in spite of the fact that what is actually on offer – realism, operationalism, epistemological anarchism, etc. – is itself the product of a competitive process.
Can institutional control of knowledge flows ever be really effective? Power holders must reckon with the fact that any given distribution of knowledge within the social system depends on how far it has been codified and abstracted. Knowledge travels more easily when it has shed data and been compressed into codes and categories. A social system's core schemata, however, the key assumptions about the world that hold it together, are usually implicitly held and only partially articulated.125 This sometimes produces a dilemma. In large, diverse groupings subject to a variety of socialization processes, social cohesiveness requires that core schemata be suitably structured for transmission to a wider audience. Recall, however, that both codification and abstraction are hazardous as they may bring to the surface latent conflicts between competing groups or inconsistencies that would remain undisturbed if left unformulated. Articulating core schemata, therefore, when tempered by the need to avoid social conflict or confusion, is likely to produce distorting oversimplifications in the form of easily remembered and uncontroversial institutional formulae. Cultural and intellectual impoverishment will then be the price paid for the pursuit of cohesiveness in societies that cannot metabolize the conflict between competing core schemata. Paradoxically, where these simplified formulae turn out to be amenable to empirical testing and refutation, the cohesiveness sought through them may be placed even further out of reach.126 The recent disintegration of formerly communist societies offers eloquent testimony on this point.
The limits to institutionalization
The dilemma of institutionalization can be summarized thus: the more people that have successfully absorbed or been socialized to a system's core schemata and values, the less will its governance be experienced by them as an external imposition. Less coercion and fewer external controls will then be required, with a consequent reduction in the costs of social exchange. In face-to-face or primary groups, socialization can take place quite unconsciously so that both the system's core values and its related institutional order can remain implicit. But as groups get larger and more diverse, and as they become more spatially scattered, interpersonal bonds loosen and face-to-face relationships give way to more anonymous transactions. Then, the tacit component present in all social communication no longer diffuses smoothly or uniformly through the social fabric.127 An uncodified residue gets left behind in any cultural transmission, imposing a margin of personal discretion on any effort at interpretation by a would-be recipient that he/she may not even be aware of. The most conformist individual is thus turned into something of an inventor regardless of how reluctant he/she may be to exercise this talent. Data loss is inherent in all transmission of information. It is indicative of entropic processes at work in the system.128 Institutions, no less than physical processes, must reckon with the second law of thermodynamics.
3.10: THE ENTROPY OF SOCIAL PROCESSES
Entropy as a social loss of memory
Institutions are a battle against entropy, against forgetting and the consequent loss of organization. Through judicious acts of codification and abstraction they structure and store knowledge that has or is deemed to have order-preserving properties. In an effort to minimize entropy production, institutions are continuously constructing or reconstructing the social order by selectively moving knowledge up and across the U-space from quadrant 1 to quadrant 3.129
Codification acts to select what can be stored and preserved, abstraction to establish how widely and usefully it can then be applied as an ordering principle. Here we are mainly concerned with the storage properties of codes and hence with codification. In discussing codification, however, we must guard against oversimplifications such as, for example, equating writing with codified communication and speaking with uncodified communication just because the first seems to have a more durable substrate: paper. Ordinarily, writing does, seem to involve a greater effort at codification than does speech, but not always. An after-dinner speech, for example, may be a far more formal affair, even when given from brief jottings on a piece of paper, than, say, the automatic writing of a surrealist like André Breton which might be totally impulsive. Or, to take a more telling example, an oral tradition is effectively an exercise in codification that does not involve the use of writing at all. Here, of course, the degree of codification can vary considerably from one tradition to the next, being a random transmission from group to group or from generation to generation in one instance, and a carefully regulated affair using special mnemonic devices and techniques in the other.
The channels available for transmission and the physical mechanisms available for the storage of information will favour particular types of code. To take one example, the quipu is used in Peru as a mnemonic device.130 It is a series of knotted cords of different colours and lengths, tied together and attached to the head-dress as a fringe. The Quipu can be read since colours, knots, lengths of string all have mnemonic significance. The line that divides a mnemonic device as an aid to human memory from an external store of knowledge as an alternative to human memory is not always a clear one. For this reason the transition from an oral to a written tradition may sometimes be imperceptible. The design of an everyday object as an act of codification, therefore, could serve to reinforce an oral or a written tradition equally well.131
If mnemonic devices are individual aids to reliability, some societies maintain the reliability of their oral traditions by social means. When the Kuba, for example, want to recite a group testimony, they set up a Kuum or secret conclave where a spokesman is appointed and the testimony rehearsed so that all are in agreement with it before it is recited in public. More coercively, in Polynesia, ritual sanctions are brought to bear on the speaker if his performance is less than word perfect. In New Zealand, not so long ago, the slightest deviation from the tradition brought instant death to its reciter – a pretty drastic response, one might think, to the threat of entropy.132
The risks incurred by the individual narrator when public control of the tradition is so punitive are such that he/she will always cautiously be confined to a minimal version of it. Yet even in the absence of punishment, according to Vansina, ‘A group testimony is always a minimum testimony.’133 Thus once more and as always, codification, this time verbal, entails a loss of data and possibly of information as well. And where the degree of social control is weakened by the availability of several traditions that the individual narrators can draw upon, personality factors will enter into both their selection and their presentation. An element of personal discretion thus pervades any description of the event, eroding the rigour of the codification process itself as well as the durability of what is subsequently stored. Entropy then wins out.
Vansina has suggested that another way of maintaining the accuracy of oral traditions is to limit their diffusion by making them the property of special groups. They then become esoteric knowledge that others may possess but are not allowed to transmit. In Rwanda, for example, certain families are entrusted with the guardianship of selected traditions in exchange for social privileges. If they want to hold on to the privileges, the chosen families must prevent the traditions from becoming public property. The link between uncontrolled diffusion and entropy has clearly been grasped here: the custodians of the Ubwiiru, the dynastic code of Rwanda, are not allowed to publish it, particularly in the Rwanda language, for fear that it might be transmitted clandestinely with who knows what additions and omissions.134
We are now in possession of an insight important to our later discussions: one way to control the diffusion of knowledge in a social system is to make it appropriable, to endow it with property rights. By establishing either exclusive or ordered access to valued resources, property rights effectively reduce the social production of entropy.135 Among Tobriand Islanders of the Pacific, a tradition can actually be bought. Thus a certain Chief Omarakana once gave food and objects of value to the descendants of a certain Tomakam in exchange for the dance, the song, and the commentary on it that they owned.136 A more familiar example drawn from our own industrial societies is the technical licensing agreement in which proprietary knowledge and know-how, subject to certain restrictions, is made available to others on a selective basis in exchange for royalty payments. Note that in both instances it is the act of codification itself which, by giving knowledge a visible structure, makes it appropriable and hence the object of a controlled diffusion. In some societies, the state itself will exercise control over the codification and diffusion of knowledge – in its day the Inca state was no less rigorous in this respect than was the former Soviet Union before perestroïka.137 In others, control will be delegated to trusted groups, custodians of a social order in which they have a stake. Yet whatever the control mechanism, and however loosely or restrictively it is applied, the transmission of socially meaningful knowledge – of clans, of family histories, of science, and of schools of thought – is institutionalized in such a way as to maintain its socially perceived accuracy and authenticity. In short, it is institutionalized so as to minimize information losses or entropy.
Entropy as a loss of socially derived meaning
Oral traditions, as might be expected, do not rank very highly in terms of accuracy and authenticity, being placed somewhere between the eyewitness account and the rumour. They are cases of weak codifications, of information diffused in face-to-face situations within comparatively small groups – more often than not pre-literate. Given the small numbers involved and their spatial contiguity, however, the socially meaningful knowledge made available to an individual can quite easily be institutionally and hierarchically regulated. By contrast, where numbers are much bigger, and constant spatial proximity less likely, the individual becomes a recipient of knowledge that flows much more anonymously, that is readily available from a variety of sources, and that is therefore much less amenable to rigorous hierarchical control. Such knowledge may well have originated on the left-hand side of Figure 3.4, but as it spreads informally through a population it becomes prey to entropic processes that sap both its accuracy and its authenticity. Any individual located to the right on the diffusion scale of Figure 3.4 – i.e., a recipient rather than a transmitter – is thus bombarded with messages of uncertain provenance that compete for his attention while often contradicting each other.
In industrialized cultures, television advertising is the paradigm example of such messages. Consciously or otherwise, the individual copes with the resulting cognitive overload firstly by downgrading the status of each message received from this source and then by becoming more selective and deliberative in his responses. In short, the individual, faced with an incipient information chaos building up on the right of the diffusion scale, is led to impose her own organization on the data through successive acts of rational choice. If on the left of the scale, the validity and utility of transmitted messages can still be underwritten through the legitimate exercise of institutional power, on the right, the individual recipient has to assess these for himself. She will be gauging message utility against a personal utility function rather than one that has been institutionally imputed to her, and the validity of the message, given its uncertain provenance, will be established with reference to its internal consistency rather than to its origins.138 Do we not have here an explanation for the greater rationality of fashion as observed by Tarde when he compared it to tradition?139
The exercise of individual rationality, as we have already seen, carries a price tag. The pluralistic competition that takes place between a variety of implicit cosmological orders or meaning systems is subversive of a single stable, taken-for-granted definition of reality. At a moderate level of intensity, such competition may in fact be highly adaptive and may facilitate the process of change and adjustment that characterizes peaceful social evolution. Beyond a certain point, however, and particularly when the framework of rules that govern the competitive process itself becomes a target for change, unbridled competition can result in a total loss of meaning, a state of normlessness that Durkheim termed anomie, and eventually in a collapse of the institutional order which such meaning underpinned.140 The social system's capacity for assimilating and accommodating new knowledge141 then becomes overwhelmed and the world ceases to make sense under any conceivable definition of the term. A frequently cited example of such a process at work is the sudden contact between either a primitive or a traditional society on the one hand, and a modern industrial society on the other. The large number of new and unfamiliar categories imposed by the latter often leads to a destabilization or an atrophying of the former,142 especially where its world view is narrow and brittle.143 The simple apodictic notions that are capable of regulating the daily flow of events in a primitive culture are rapidly submerged in the complex currents and whirlpools that shape an industrial one.
Atrophy is one possible outcome of such exposure to overwhelming change: a ready example is provided by the lethargic and aimless existence on offer today in the residual cultural spaces of an American Indian reservation. In societies that have undergone some evolution, but which have done so within a social order moulded by tradition, however, destabilization imported from outside will be fought and might provoke a backlash. Islamic fundamentalism – a search for simple original virtues through which to combat the evils of industrialism and pluralism – here readily comes to mind as a contemporary example.
What accounts for the institutional fragility of societies whose traditions have proved so resilient in the face of nature alone? Berger and Luckmann have argued that the hyperstable order of traditional society can lead to a form of reification in which ‘the objectivated world loses its comprehensibility as a human enterprise and becomes fixated as a non-human, non-humanizable, inert facticity’.144 Imperceptibly, the abstract world of institutions merges with the concrete world of nature to occult the mediating role of human consciousness; Popper's W2 and our E-space as a whole either are then lost from view or their existence is denied altogether. In effect, no distinction is allowed between Popper's W1 and his W3; reality is seamlessly one. Pluralistic competition, when it hits these societies, then tears away the institutional veil that has been so tightly drawn over things, to reveal the frail contingent nature of their core premises – a traumatic experience for any culture that lacks the psychic resources to adjust.
Fighting entropy through institutionalization
If entropy erodes a society's sense of ontological security,145 institutionalization aims to restore it. The scope for stable institutionalization in a given society, however, ultimately depends both on how widely shared its relevance structures turn out to be and on whether these are sufficiently broadly defined – i.e., abstract – to accommodate variety and change in concrete instances. It is variety and change that stimulate the conscious search processes through which W2 is explicitly brought into play as a mediating agency between W1 and W3. Note, however, that sharing relevance structures goes beyond the simple sharing of knowledge; it entails in addition a common valuation of the knowledge base in areas where it matters. The particular social conditions under which the diffusion of knowledge effectively occurs must therefore be well understood.
In this section and the ones that have preceded it, we have emphasized that the potential utility of knowledge acts as a stimulus to its diffusion and its subsequent institutionalization. The diffusion knowledge itself thus acts as a powerful antidote to entropy increases within a system. Effective transmission for this purpose can be thought of as overcoming both level B and level C communication problems in Shannon's terminology – that is, ensuring that messages firstly convey their intended meaning and secondly lead to desirable patterns of behaviour. Stated otherwise, for effective communication to occur there must be some minimal alignment of individual E-spaces. The technical problems associated with information transmission itself, however – Shannon's level A – must not be neglected. Their resolution makes a crucial contribution to the successful diffusion of knowledge and hence to the process of social evolution. Shannon presented level A communication problems as being wholly independent of the other two levels, its own influence on them being unidirectional. Shannon was concerned with communicative efficiency: how best to transmit a message once it has been formulated. Yet one does not typically proceed by first formulating a message and then working out how to transmit it. Message formulation and transmission in most situations constitute a single project. Technical communication problems therefore can rarely be addressed in isolation as they are by the communications engineer, so that Shannon's three problem levels must be approached holistically. With that proviso we focus in the next section on some of the technical level issues raised by information sharing.
Codification and diffusion: some propositions
One tenet basic to the theoretical perspective developed in this book is that the codification and diffusibility of information are systematically related. It is our contention that the relationship is a fundamental one – no less so than that which obtains between abstraction and diffusibility – with wide implications for both psychological and sociological processes. Although for expository purposes the diffusion properties of abstraction and codification have been treated separately, in practice, as was shown in Chapter 2, the ability to abstract and to codify are intimately related. Simplifying somewhat we might say that if abstraction as a cognitive strategy economizes on content, codification economizes on form. Economies of the one, sooner or later, lead to economies of the other.
Stated algebraically, our proposition is that:
D = f(C)
Our formula thus brings together the codification and the diffusion scales already discussed in earlier sections into an orthogonal relationship where D is the percentage of a given population of data processors that can be reached in a given unit of time with a particular message, and C is the degree of codification of a unit of length of that message, defined as the reciprocal of the number of bits of information required to transmit it. At the lower end point of the codification scale, for example, the number of bits required to transmit a unit message becomes infinite and we thus obtain:
Here no conceivable structuring of the data is possible and not even scanning strategies can help; the data processing apparatus simply becomes overwhelmed by the sheer volume of data to be processed and data just runs off it. We are in the domain of the ineffable, of zero codification. At the other end point of the codification scale, by contrast, the information content of a given unit message can be captured by a single bit so that:
No further coding of this particular message is possible: its data has been reduced to its theoretical minimum.
The function f is derived from the specific characteristics of the various communication elements that link the various members of the diffusion population into a complex communication system: the number of channels available between any two links, channel capacity, channel noise, the nature of transmitters and receivers available, etc. – in short, all the elements that make up a communication system as described by Shannon and Weaver146 and that affect the choice of code and the time incurred by coding. The choice of system itself will reflect broader constraints such as the spatial distribution of the target population, the communication technologies available, their cost, and so on.147
A schematic description of the relationship between codification and diffusion is given in Figure 3.5.148 Clearly, many different curves could be drawn to reflect a more realistic variety of communication situations and most probably few of them would be as smoothly drawn as the one shown. What they will all demonstrate, we nevertheless hypothesize, is that if abstraction and codification, working together, are the means available to us for overcoming constraints on cognition, then codification is also the activity through which we overcome physical and energetic constraints imposed on acts of communication by dint of their occurrence in time and space. The extensive diffusion today of ever-greater volumes of codified data to ever-larger and more distant populations constitutes prima facie evidence in support of our claim.
Figure 3.5 The codification and diffusion of knowledge
Put thus, the proposition presented above has an almost embarrassing simplicity: knowledge that cannot be formalized in some way diffuses more slowly and less extensively than knowledge which can be. Such simplicity, as we shall presently see, is deceptive. To gain a more intuitive feel for the proposition's deeper meaning, let us start by dichotomizing our two dimensions as follows:
The codification dimension
1Uncodified knowledge: knowledge that cannot be captured in writing or stored without losing the essentials of the experience it relates to – i.e., of the smile on your child's face; of a sunrise on Mount Taishan; of riding a bicycle; of playing a guitar; of a headache.
2Codified knowledge: knowledge that can be stored or put down in writing without incurring undue losses of information – i.e., stock market prices; the Tudor Constitution; the chemical formula for benzene; a postal address.
The diffusion dimension
1Undiffused knowledge: knowledge that stays locked inside one's head whether because it is hard to articulate or because one decides to keep it there – i.e., childhood memories; military secrets; sexual fantasies; skilled performance; a toothache.
2Diffused knowledge: knowledge that is shared with others – i.e., the dates of historical events; a concert; a radio or television broadcast; Mae West's published biography; the law of contract; Boyle's law; a good laugh.
A typology of knowledge
We can intuitively assign the items of knowledge listed under each dimension to one of the four quadrants of Figure 3.6. Most entries will prove to be quite straightforward: the personal experience of a sunrise on Mount Taishan, for example, is both hard to codify and hard to share with those not present; knowledge of stock market prices on the other hand is at the same time easy to set down on paper and easy to diffuse. Some entries, however, will be more troublesome. What, for example, are we to make of a concert? It does not easily fit into a dichotomized structure. The experience of a concert might be hard to codify and to share with those not present, but its production - i.e., following and responding to a written score – presents fewer problems in this respect and admits of a limited diffusion, albeit one that until the advent of broadcasting was spatially restricted to an audience sitting in an auditorium. If we make due allowance for the dangers of oversimplification, the result of this exercise is the fourfold classification of knowledge shown in Figure 3.6. Each category is now discussed in turn.
Figure 3.6 A typology of knowledge (2)
Public knowledge
Public knowledge is what most commonly passes as knowledge in society. It is structured, tested, recorded in textbooks, learned journals, and other publications.149 Less loftily it may also appear as product information on beer cans or cornflakes packets, as television advertisements, as scandalous publications, and as railway timetables. In the case of scientific knowledge, gaining access to this quadrant for the purpose of diffusing new knowledge is subject to a certain number of restrictions, but once it is in there its diffusion is hard to control. As with other forms of public knowledge, people are free to take it or leave it, distort it, repackage it, add to it, and so on; it becomes a case of caveat emptor.
A key point about communication patterns in this quadrant is that where the communicating population is large, people often transact anonymously. Transmitters and receivers may never meet, and even if the data they deal in is hard and codified, unless it has been branded in such a way that its source can be identified it will not be easy to authenticate. The branding of an item of public knowledge invests it with authority and improved its chances of adoption and hence of survival in the competitive marketplace.
Even where the diffusion of public knowledge involves face-to-face interaction it tends to be mostly impersonal. Few of the customers who buy their evening paper from a news-stand on the way home from work, for example, ever get beyond a brief verbal exchange with the vendor. Some prior familiarity with the paper itself and a knowledge of its price will usually suffice to keep the transaction brief and virtually silent.
Public knowledge by its very diffusibility is detachable from its origins and over time gains a certain facticity. It is, if anything, more attached to the coding procedures that gave it structure and form than to the situational context from which it emerged. Well-chosen codes bestow upon their data valuable powers of combination, allowing them to form novel patterns either on their own or by linking up with existing items of codified knowledge. In this way public knowledge slowly gets woven into a dense impermeable tissue of facts, categories, and concepts that may subsequently prove hard to modify – i.e., over time it acquires inertia.
Common-sense knowledge
Much less codified than public knowledge but no less widespread is common-sense knowledge. Common sense is acquired more slowly than specific items of codified knowledge and in a different way: public knowledge such as textbooks may provide the raw materials for its development but what gets internalized as common sense largely reflects a person's social situation and the power that others have in orienting what he learns about the world and how he makes sense of it. Common-sense knowledge thus has a more contingent feel to it than public knowledge. A person's stock of common sense is acquired gradually over a lifetime through an idiosyncratic distillation of personal learning experiences and face-to-face encounters that are initially confined to the family circle but that subsequently broaden out to cover peers and other members of the person's culture.150 Common sense is context dependent and its exercise cannot be divorced from the social value and belief system that moulds the pattern of preferences with reference to which consistency of behaviour will be judged. What will appear self-evident to a family of Gansu subsistence farmers may leave a group of Wall Street stockbrokers scratching their heads. It is not, however, necessarily to indulge in cultural relativism to argue that although each may achieve a measure of internal consistency, some forms of common sense will prove more viable than others since there may exist superordinate grounds for preferring one value system to another. It may for example be able to handle a greater variety of contexts without loss of effectiveness or consistency.151
In dealing with common sense a dichotomized diffusion scale is somewhat misleading, since not only does this kind of knowledge diffuse inter-personally much more slowly than codified knowledge – this is in line with our main proposition – but because of this it also tends to spread only partially through a given population. Even within a single culture, therefore, subgroups will crystallize around implicit beliefs and value systems that may never diffuse much beyond their boundaries. The resulting discontinuities in the social structure will further limit the diffusion of what passes for common sense in different groups so that only those elements of it which are derivable from a culture's core values, and that are subscribed to by the population as a whole, will be widely shared.
Personal knowledge
Personal knowledge shares with common sense a diffuse and ambiguous quality but it is by nature much more idiosyncratic.152 Both are hard to articulate and are usually communicated implicitly, often by force of example. Personal knowledge, however, even in face-to-face situations, is less accessible than common sense. It has almost by definition insufficient overlap with the experience of others – this also applies to the primary group – to create a shared context suitable for discourse. And with no shared context, personal knowledge cannot compensate for the absence of a shared code that would otherwise be needed for its transmission. Polanyi makes the point that personal knowledge is not necessarily subjective knowledge, arguing that ‘the personal is neither subjective nor objective. In so far as the personal submits to requirements acknowledged by itself as independent of itself, it is not subjective; but in so far as it is an action guided by individual passions, it is not objective either. It transcends the disjunction between subjective and objective.’153 Personal knowledge that gradually does diffuse towards the right-hand quadrants of Figure 3.6, and hence becomes shared, acquires a certain intersubjective validity and freedom from the purely subjective organization of the individual experience. It becomes Objectivated’ in the form of traditions, skills, recipes for conduct, etc.,154 and might subsequently become the common possession of a group or society. Of course, none of this transmission need occur at an explicit level, at least in its early stages. As diffusion extends over larger reaches of space and time, however, some of the more recurrent features of a personal experience, now increasingly widely shared, may become candidates for some form of codification. Those that are not will remain ephemeral and are ultimately destined for extinction.
Where personal experience encounters discernible regularities in the flux of events that impinge upon it, stable patterns may emerge from it and generalizations may suggest themselves. We described this in Chapter 2 as a process of codification and abstraction that moves an individual through her E-space and makes her experience increasingly intelligible and articulate.
Typically, in structuring the E-space thus, an individual will draw upon a repertoire of existing codes and concepts made available by her culture: the individual inherits ways of knowing that over time help to align her personal experiences with those of ancestors, contemporaries, and descendants and thus renders discourse possible. Situations may arise, however, in which the cultural repertoire of codes available to an individual fail her, denying or distorting personal experiences that are irreducibly real and in urgent need of interpretation. Through a process of trial and error, the individual may then be led to devise a personal coding scheme that helps to make sense of her situation, even if a chosen scheme socially isolates her and causes the individual to be branded as a deviant.155
Where the experience thus codified is totally idiosyncratic and unique, it will lack the degree of overlap necessary to make it relevant or interesting to contemporaries and will not justify any effort on their part at mastering the new code. The individual in possession of such a code has, in effect, created a private language for himself, one which may facilitate autocommunication but little else. Where some convergence exists between an individual's idiosyncratic experience and that of others, however, pressures may grow either to diffuse the newly articulated knowledge and codes, or to block their diffusion. Blocking will occur where receivers perceive them as a threat; it will also occur where transmitters deem them too valuable to just give away.
We shall call the structured knowledge that emerges from idiosyncratic coding processes ‘proprietary knowledge’: once codified it becomes in effect technically diffusible but while it continues to reside in a single mind it cannot easily be appropriated by others. Proprietary knowledge, of course, is not necessarily useful knowledge: the codes and solution devised by an individual to make sense of a unique situation may have little general application outside it or may turn out to be faulty or distorting – i.e., poorly formulated perceptual or conceptual hypotheses.
But where such knowledge does prove to be useful, however, it may now be traded and bargained over. Being as yet undiffused it has a degree of scarcity which when combined with its utility gives it value.156 Diffusing it into the right-hand quadrants of Figure 3.6 erodes its scarcity and hence its value. Diffusion barriers will therefore often be erected by possessors of proprietary knowledge between the top left-hand and the top right-hand quadrants in order to slow down or prevent uncontrolled leakages of potentially valuable information.157
The codification and diffusion dimensions of Figure 3.5 create a space which allows us to explore the way that different types of information and knowledge are structured and shared within a given population. The structuring of knowledge expresses cognitive activities that take place in individual E-spaces, an interplay of abstraction and formalization that gets registered in the figure as one moves up the codification scale. Chapter 6 will develop the proposition that information structuring and sharing are the defining attributes of a cultural process. For this reason we propose to label the space of Figure 3.5 a culture space or a C-space for short.
A C-space can be used at many levels. In so far as an individual, for example, is a recipient of existing and available knowledge as well as a generator of new knowledge, he/she can be profiled in the C-space drawing on the typology presented in Figure 3.6. His cognitive assets might first be located in his personal E-space and then subsequently in a C-space where their scarcity value can be gauged by asking ‘what does he know that others don't know?’ Others’, of course, need specifying. This is done by defining the population to be placed on the diffusion scale – a village, the employees of a firm, customers, a nation-state, etc. Scarcity value is then measured with reference to this specific population.
Used thus, the E-, U-, and C-spaces combine into a single three-dimensional representation to give us the cognitive ‘signature’ of an individual, a small group, or a larger population, and point to the probable strategies they will respectively adopt in receiving, processing, and transmitting information. Strategies for the reception and processing of information were discussed in Chapter 2 under the heading of learning styles. We must now examine more closely the strategies available for the transmission of information.158
Sharing of codes versus sharing of context
Providing suitable communication channels were available, in theory any message, no matter how complex and consuming of data, could be technically transmitted between any two points aligned horizontally with each other in the C-space – i.e., operating at similar levels of codification. The time this would require, however, the degree of attentiveness it would impose on both senders and receivers, and above all the presence of noise all conspire to set practical limits on what can effectively be communicated. Intelligible communication is no free lunch.159
Since no absolute structural distinction can be drawn between noise and signal, the presence of noise in the C-space is problematic. According to Moles, noise can be thought of as a signal that a sender does not want to transmit or that a receiver does not want to receive.160 The respective intentions of senders and receivers can then act as guides to help us distinguish noise from information. Yet, in the lower part of the C-space, intentions may be quite ambiguous or vague; consequently, the presence of noise there will be proportionately more damaging to effective communication than in the upper part of the space. Moles argues that The only way to reduce noise, is to reduce the channel's capacity, the extent of the repertoire of elements that it can convey ... The choice of elements is reduced [and] we particularize, a priori, the nature of the signals which we want to amplify or receive ... What is gained in sensitivity is lost in the variety of elements ... We have derived a principle limiting the amount of information which can be received from the external world ... As a consequence, in order to increase sensitivity indefinitely, it is necessary to know more and more about the nature (frequency) of the message to be received.’161
Figure 3.7 The effect of noise on the diffusion curve
Moles's prescription for dealing with noise – a reduction of channel capacity – leads us in two directions. The first is towards a greater degree of codification of messages passing through the channel in order to make more efficient use of the channel capacity that is left; this, of course, requires a prior sharing of codes between senders and receivers. The second holds the level of codification constant and leads instead towards a greater prior sharing of context between the communicating parties. In both instances the a priori knowledge shared by the parties must be increased. But whereas in the first case anonymity and distance can be maintained, the second entails a much greater degree of face-to-face involvement: the mutual adjustment and alignment of expectations that the sharing of context makes possible requires a high degree of co-presence. This, of course, limits the numbers that can be involved in a communicative exchange so that sometimes the best results are obtained by mixing the two approaches. The example given by Edward Hall of an airline pilot communicating with air traffic control (section 3.7) combines the sharing of codes with some sharing of context. For the latter to occur it is not necessary that every airline pilot develop a face-to-face relationship with every air traffic controller he will have to deal with; it is merely required that each group be socialized to the other's values and ways of thinking concerning those tasks on which they have to collaborate and get on to the same wavelength.
Noise shifts the diffusion curve of Figure 3.5 leftward from AA’ to BB’ as in Figure 3.7. Responding to noise by sharing codes or by sharing context has the effect of placing the receiver on different points of the now shifted curve. One can either increase the level of codification of a message and hold the size of a target audience constant, or one can reduce the size of a target audience so as to handle noise-related ambiguities interpersonally.
Increasing the level of a codification of a message entails a sacrifice at least of data and possibly also of information, both of which may lead to communicative distortions.162 Nuances that were present in the sender's mind have to be expunged from the message in the interest of clarity and efficiency; subtleties evocative of latent or alternative possibilities are bracketed out; personalized qualitative elements that express authorship, style, and preference must now give way to more bland and neutral forms.
Yet holding the degree of codification constant also has its problems. The threat of message distortions and information losses imposes co-presence and thus limits the size of the audience which can participate in more ambiguous transactions. One of the great advantages of face-to-face communication is that it is multichannel and, where required, it can therefore selectively increase the level of redundancy in order to compensate for the presence of noise. The meaning carried by a tone of voice, for example, can be reinforced by the raising of an eyebrow or a hand. Whether harsh words are pronounced in earnest or merely in jest can usually be intuited from, say, the nature of the accompanying smile. Skilfully correlated signals, therefore, emanating from different sensory channels, can be used to amplify or attenuate the central message. This requires physical proximity, however, and even then it can never by itself eliminate the ambiguity that some irreducible lack of shared codes – behavioural as well as verbal – imposes upon all social exchange. There will always be a sense in which the sender ‘knows’ more than the receiver, retaining a ‘tacit coefficient’163 which he is unable to articulate no matter how much he may wish to. Indeed, Freudian theory holds that the sender himself never really ‘knows’ how much he knows, much of his knowledge being submerged within the depths of his unconscious, and hence beyond the reach of any degree of articulation. Thus the communicative act, in an absolute sense, is always incomplete, even when it is reflexive, and thus addressed to oneself.164
Branding as the creation of trust
Asymmetries in the knowledge held respectively by sender and receiver, whether they are intended or not, pose the problem of trust and expectations as a determinant of effective communications. Where the message is uncodified, trust has to reside in the quality of the personal relationship that binds the parties through shared values and expectations rather than in the intrinsic plausibility of the message or other features of the communicative situation – i.e., it resides in their ability to get attuned to each other, to get on to the same ‘wavelength’. Trust, in short, reflects a prior investment in the communication nexus, undertaken to counter the anticipated presence of noise in the channel and to speed up the pace at which ambiguous or uncertain (i.e., uncodified) messages can be transmitted. Trust requires ‘high context’ in Hall's sense of the term;165 it allows the use of a shorthand – a form of coding based on shared contextual knowledge – that reflects the parties’ attunement to each other, what Giddens describes a mutual knowledge.166
Yet trust is also called for in dealing with more codified messages where these are anonymous. For what, after all, might be their provenance? And what degree of reliability can be placed upon them? Being well structured they may be quite unambiguous but still designed to mislead. Where a sender cannot be known to a receiver, it is the credibility of the channel itself that has to be built up. It then becomes the focus of trust-building efforts – let us call them branding efforts. Brands confer authorship, and hence identity, on messages that one might otherwise be led to disregard for want of trust. In the world of commerce, brands aim to shape consumer expectations in positive ways and to build up goodwill towards a message source.
In the world of commerce, however, brands authenticate messages that refer to a limited set of concrete goods or services, W1 objects that can be eventually sampled and experienced directly. Risks, although certainly present, are confined to specific members of the set. Any attempt to extend unduly the class of objects covered by a brand – known in the jargon of advertisers as ‘brand stretching’ – sooner or later weakens its credibility.167
W3 objects, on the other hand, abstractions that can find their way into myriad W1 situations, have much more potency. And being consumed indirectly – through the W1 objects they embed themselves in – they can be ‘brand stretched’ to a far greater extent than specific consumer products or services can be.168 The potential audience for such W3 products is consequently much larger and is often in need of some protection against either fraudulent or questionable message sources and the damage that these can inflict. In the case of the most abstract W3 objects, the products of science, the responsibility for protecting the integrity of the ‘brand’ is entrusted to the scientific community as a whole; it regulates the diffusion of new knowledge produced by its members through institutionalized publication procedures that are designed to weed out unsustainable claims.169
IT and the extension of strategic options
To summarize thus far: prior investments, both in communicative relations between social actors and in a given communication infrastructure and its associated technologies, will give the diffusion curve schematically depicted in Figure 3.5 a specific shape that will reflect, in addition to such investments themselves, the characteristics of a target population such as its size, its spatiotemporal distribution, its coding and abstracting skills, and so on. Effective communication strategies will be those that take such characteristics into account when tailoring a given message's degree of codification and abstraction to the possibilities and expectations of the chosen audience.
Our strategic options in matters of communication are today constantly being extended and amplified by the evolution of new information and communication technologies. Certain uncodifiable qualities of face-to-face interactions, for instance, can now be diffused more widely through the spread of videoconferencing which thus confers on selected events an immediacy that in earlier times was totally unavailable outside the here-and-now. The effects of such developments should not be exaggerated, however: a televised concert and a live one are clearly not the same thing, and the prices that concert-goers are willing to pay to experience the latter directly bear witness to the difference.170 Yet with ever-accelerating technical change, one could hypothesize a long-term shift of the diffusion curve towards the right in the C-space towards CC’ as depicted in Figure 3.8. The shift registers the ability of emergent telecommunication and computer technologies to handle ever-larger volumes of data, and hence, in many instances, also a diminishing need either for the communicative efficiencies offered by increases in codification171 or for restrictions in audience size, each being in part a response to the presence of noise in the system. Figures 3.7 and 3.8 are in effect mirror images of each other. The leftward shift of the diffusion curve shown in Figure 3.7 amounts to a restriction on communicative options; the shift to the right in Figure 3.8, by contrast, amounts to an extension of communicative options.
Figure 3.8 The effect of information technology on the diffusion curve
The diffusion curve in the C-space, however, reflects little more than the technical characteristics of a communication system, an ability to overcome Shannon's level A problems at different levels of codification as measured by the number of people that can be reached. It describes the diffusibility of information rather than the level of diffusion actually achieved. It does not ask, for example, how a receiver came to share codes or contexts in common with a given sender. This is Shannon's level B problem – the semantic problem. It is a far from trivial problem, of course, since for meaning to be shared by communicating parties, not only must the same codes be learnt by each but they must also be given substantially the same referents; that is, they must be contextualized in a similar way. A lack of prior shared contexts, however, being symptomatic of the impersonal and codified communication nexus, makes it less likely that codes and referents will overlap to any great extent in any given exchange.
Restricted versus elaborated codes
In cases where the skilled use of a code promises access to knowledge with a high degree of utility, it will often justify an individual's prior investment in mastering the code – as for instance in technical or scientific training. Any differences in interpretation of a common pool of knowledge held by communicating parties are then brought into mutual alignment by coding conventions that facilitate the communication process. Yet in so far as a code's potential utility is tied to the degree of generality that it can achieve, the conventions that regulate its applications and its use in communication will crystallize in the north-east region of the E-space where knowledge is abstract as well as codified and can only be acquired by those willing to invest extensively in the appropriate type of theorizing.
But what is ‘appropriate theorizing’? In highly codified and abstract fields of knowledge such as modern mathematics, for example, Shannon's technical level problems present few challenges to effective transmission; yet those at the semantic level abound. If today no single mathematician is able to master more than a tiny fraction of the discipline's theoretical body of knowledge, how do mathematicians working in different subdisciplines with different symbolic codes and concepts ever manage to communicate?
In asking how different specializations communicate across conceptual and coding barriers, we are effectively invoking the distinction drawn by Bernstein between ‘restricted’ and ‘elaborated’ codes. For Bernstein a restricted code operates on the basis of a shared context and limited codification. The implicit nature of the code, and above all the requirements for a shared context, act to restrict the size of the linguistic community with which one is able to interact. This was the problem that confronted the working-class children that Bernstein studied. We would locate them in the south-east corner of the C-space. The more explicitly elaborated code that middle-class children were taught to use, on the other hand – these were also studied by Bernstein – placed them in effect higher up the C-space and thus served to extend the size of the linguistic community with which they could interact. An increase in codification here reduced the requirement for shared context.172
But does not the specialized mathematician also work with a restricted code, one that by its very abstraction and codification limits the number of people that have access to it? The mathematician incurs isolation by moving even further up the codification scale than Bernstein's middle-class children do. Viewed purely technically, the symbols he/she manipulates could in principle be transmitted faster and hence further in a given unit of time than anything verbally articulated by middle-class children anywhere – electronic networking, for example, can today link up scientific databases instantaneously on a global basis. Yet the investment in time and effort required to master the coding conventions used in different subfields of mathematics, given their high degree of abstraction, limits the number of people willing to master the code to a tiny fraction of those who could effectively be reached communicatively through it. It is in this sense that the code is ‘restricted’ and the restriction operates in a quite different way from what is described by Bernstein.173
The semantic curve
The mathematician has a communication problem at Shannon's level B, the semantic level. Figure 3.9 schematizes it as a semantic curve. The curve gives us the percentage of a given population potentially capable of understanding a given unit of message in a given unit of time. Here codification and diffusion are inversely related: the physical diffusibility of information as one moves up the space may still go on increasing but effective diffusion, taken in the sense of a message communicated, received, and understood, will actually reduce.
Figure 3.9 The semantic curve in the C-space
What accounts for this inverse relationship? Recall from Chapter 2 that abstraction is facilitated by codification and that they mutually reinforce each other. The relationship between them was depicted in Figure 2.5. Abstraction, however, as we have seen, can be performed in any number of ways and there is no assurance that what these will mean for different individuals can be brought into alignment unless it be by learnt coding conventions requiring time and effort to master. Our new semantic curve, now projected onto the U-space in Figure 3.10, also depicts an inverse relationship between the degree of abstraction of an item of knowledge and the size of the population that can master it. The curve strongly suggests that a negative utility or cost is incurred by individuals investing in abstract codes.
For anyone contemplating investment in the acquisition of abstract knowledge or coding skills, therefore, something must compensate him for the costs and effort involved. That something turns out to be whatever appropriable value is realized by the joint utility and scarcity of what is acquired. Utility gains are achieved firstly by moves towards abstraction on account of an increase in the generality of the knowledge thus obtained, and secondly by moves towards codification which give form and stability to such knowledge and render it manipulable. Scarcity, by contrast, is secured by the limited number of people willing to cognitively invest.
Many considerations dictate an individual's cognitive investment choices. Few abstract coding schemes, for example, can be acquired and used incrementally; they have to be mastered as a whole. One does not decide to learn only the first twelve letters of the alphabet just to see how things go; for the attainment of literacy, it is twenty-six letters or nothing. Many technical subjects likewise require the mastery of a whole corpus of codified knowledge for even a minimal proficiency in their use. Then there is the question of variations in individual aptitude. Our discussion of learning styles in Chapter 2 pointed to considerable individual differences in cognitive preferences and skills in handling abstraction. Any move towards it will clearly be experienced as more costly by some than by others. Finally, profitable opportunities for the exploitation of abstract codes, once mastered, are not evenly distributed across individuals. For every ‘technical expert third class’ working at the Berne Patent Office, for example, who makes it to the Princeton Institute of Advanced Studies – the case of Einstein – there will be countless dustmen and ticket collectors who will not think it particularly worth while to invest themselves in a mastery of the Kaluza-Klein theory or Fermi-Dirac statistics. The cost-benefit equation they face is stacked firmly against it.
Figure 3.10 The semantic curve in the U-space
Value in the U-space is the outcome of a tension created between the two curves A and B shown in Figure 3.11. The first, A, measures the utility of knowledge as a function of the number of people for which it has potential relevance – whether or not they are themselves capable of understanding such knowledge. The greater its degree of abstraction, the larger the number of situations in which a given item of knowledge will have potential utility. The second curve, B, is the semantic curve of Figure 3.10: it traces the increasing scarcity of knowledge as a function both of its degree of abstraction and of the risky prior investments in the mastery of codes it consequently requires. Curve A is a demand curve for knowledge, curve B a supply curve. The greater the horizontal gap between them, the higher the monopoly rents that can be extracted from the knowledge one possesses.
Figure 3.11 Value in the C-space
Information strategies in the data field
The three-dimensional integration of the E-, U-, and C-spaces shown in Figure 3.12 brings into focus forces that act upon human information processing and transmission activities. We can think of these forces as creating a data field,174 a region in which an information-charged item of data exerts its influence and shapes the information environment that an individual confronts. The data field itself, however, forms part of the individual's physical environment. It consists of low-level energy which, being registered by the individual as data rather than mechanical energy, has the capacity of acting informationally. The data field is not static but evolves over time as a function of:
•the development of new codes assignable to different points along the codification dimension;
•the development of new communication infrastructures that increase the size of the audience reachable at different levels of codification across space and time;
•the size of the semantically competent audience created through investments in education;
•the development of new abstract categories assignable to different points along the abstraction dimension;
•changes in the size of the audience for which an item of knowledge has potential relevance;
•the creation of barriers to communication, deliberately or otherwise.
Figure 3.12 Integrating the E-, U-, and C-spaces
An individual's life prospects turn on his ability to respond strategically to the opportunity and threats that operate in his information environment. He will need to develop an effective information strategy, one that creates value for him by making effective use of his location as a data recipient, processor, and transmitter in an evolving data field. This requires some knowledge of the dynamics that regulate the behaviour of the field and some sense of how to invest scarce data processing and communicative resources so as to harness the field's potential.
Since we are asserting that a grasp of the field's dynamic behaviour is a prerequisite for the elaboration of an effective information strategy, we make it the subject of the next chapter.
The main propositions derivable from this chapter will now be briefly presented and some of their implications touched upon.
1The structuring and sharing of information are related: structuring, taken as any move through the E-space towards greater codification and abstraction, facilitates movement in both the U-space and the C-space and is in turn affected by these. The C-space relates both the diffusibility and the actual diffusion of information in a target population to how far it has been codified as well as to the prior mastery of codes; the diffusion and semantic curves respectively offer a graphic representation of Shannon's technical-and semantic-level communication problems. The U-space links the diffusion potential of information to its degree of abstraction, since together with codification, this affects the number of situations for which it has potential relevance or utility. It thus allows us to explore Shannon's pragmatic level communication problem.
2Where abstraction calls for investments in specialized codes and coding conventions it becomes a source of information scarcity since fewer people will be, or perceive themselves to be, in a position to benefit from the cognitive investments required. Moves up the U-space towards greater abstraction simultaneously increase the utility of information and its scarcity. It achieves the first by increasing the size of the population for which it has potential relevance (whether this is consciously appreciated by this population or not); it brings about the second by increasing the size of the cognitive investment that has to be made and hence reducing the number of those willing to make it. Only when utility and scarcity are joined in this way does information actually acquire economic value.
3The value of information is precarious and under a constant threat of erosion. Codification and abstraction simultaneously facilitate appropriability and diffusibility. It is this dual effect of information structuring that distinguishes its economic properties from those of physical goods. Unlike the latter, information goods cannot be spatiotemporally confined. Property rights in information, therefore, cannot enjoy the degree of institutional stability required for the effective operation of market equilibrating processes. The economic value of information must in some fundamental sense remain indeterminate.
4The E-, U-, and C-spaces described in this and the previous chapter reveal the forces – resolvable into three dimensions – at work in the data field. The field evolves over time, creating and distributing new information and gradually eroding or eliminating old information. Individuals create value for themselves by judiciously exploiting the forces at work in the field; for example, by generating abstract and hence potentially useful information and then by successfully keeping it local and scarce. But in doing so they incur costs measured in time and energy expended. An effective information strategy, then, is one that creates value while economizing on scarce cognitive and communicative resources. Such a strategy, however, to be effective, requires an understanding of how the field evolves over time and of how its dynamic behaviour can be exploited – the subject of the next chapter.
Implications
In one respect, the data field stands in a similar relationship to neoclassical economics as Maxwell's electromagnetic field a century ago stood in relation to Newtonian mechanics.175 For Newton, space was empty, infinite, and timeless, a neutral medium in which the only reality was the encounter between pieces of inert matter regulated by the laws of celestial mechanics. In this space, continuity and equilibrium were taken as norms and change was perceived as a disturbance that had to be explained. James Clerk Maxwell, by injecting energy into empty space, imparted a structure and a dynamic to the latter that prepared the ground for the subsequent development both of the special and general theory of relativity and of the quantum theory. Suddenly it was stability and equilibrium that had to be explained, and change – increasingly of the evolutionary kind – that now constituted the norm.
Newtonian mechanics was not thereby done away with, far from it; but it was seen to explain less than it had hitherto claimed. Building upon Galileo's earthbound experiments with falling bodies it had retained an anthropocentric flavour that confined its cosmological possibilities to what could be intuited directly through the five senses. Maxwell's electromagnetic field, by contrast, was beyond the reach of naive sensory intuition; it thus represented a quantum leap in the degree of abstraction that would henceforth be expected in the thinking of physicists.
In coming to terms with data and information, economics is having to move along a path similar to the one taken by the physicists over a century ago. Neoclassical economists, however, set out on their journey with a particular handicap, since neither physical space nor historical time form any part of their discipline's paradigmatic core: economic activity is taken as occurring at a single point in space-time and for that reason neither evolution nor the differential distribution of information to spatiotemporally situated actors is registered as a problem: by assumption, information is ubiquitous. It follows, therefore, that while such information can have utility, not being scarce, it can have no economic value; it must therefore be considered to be what economists call a public good,176 one whose consumption by one party does not increase its scarcity for other parties. Where information scarcity does occur – and economists are the first to recognize that in the real world, information scarcities abound and information does indeed have real value -it is either artificially engineered as in the patent system, or it is considered to be no more than a temporary hitch on the way to information ubiquity. The first case leads to a conceptualization of information as an object of exchange, the second to a conceptualization of information as a support for exchange. Clearly, information cannot be both at the same time. Only by restoring a spatiotemporal dimension to information processes can these conflicting conceptualizations be reconciled.
Historical space-time entered physics through the second law of thermodynamics and made its appearance in economics firstly in the writings of Marx and then later through the German Historical School, the Austrian School, and the Institutionalists.177 Marx himself regarded information scarcity as nothing more than an artificial contrivance incapable of generating real value. Value creation was the task of labour conceived as a pure energy system. Information utilities might be embodied in labour – Marx was not about to deny that brains as well as brawn were necessary in the new industrial society – but they reflected the ‘socially necessary labour time’ required to produce a good and were decreed to be a common inheritance of mankind.178 Pay differentials between skilled and unskilled labour were to be accounted for by the efforts expended by labour in acquiring specialized knowledge and information, not by the relative scarcity of those once acquired. Both the information embodied in past labour and that incorporated in new technology were by implication public goods made artificially appropriable by capitalistic property relations; they did not participate, therefore, in Marx's theorizing on historical time.
If Marx in the last century stressed the irreversible temporal aspects of economic processes,179 the Austrians in the first half of this one also attended to the spatial ones. Friedrich Hayek in particular repeatedly emphasized that all economic knowledge is at bottom local knowledge and that any plausible explanation of economic coordination must reckon with this fact.180 In the past three decades something close to a Pauline conversion has taken place in economics with respect to the role attributed to information in economic activity, professional interest focusing mainly on the way it gets distributed across economic agents,181 on its evolutionary possibilities,182 and on the institutional implications derivable from an information perspective.183
Nevertheless, a unitary view of information as both a sociological and a physical phenomenon is still lacking. Economists, for example, are often quite happy to conflate data, information, and knowledge in their theoretical discussions. Take Hayek in his discussion of the problems of knowledge in society:184 he argues that all knowledge is local. We have to agree with him since knowledge is located in individual E-spaces. But are data and information local in the same way that knowledge is? And to the same extent? Does not Hayek's stress on methodological individualism, by keeping knowledge strictly local, in fact reveal a pre-Newtonian aversion to information-based action-at-a-distance and to the emergent properties of a data field as it moves towards greater abstraction and codification? And is not the consequence of this localism – in effect an epistemological predilection for the concrete over the abstract – that organization and self-organization must be treated as epiphenomenal, thus robbing the information perspective of its paradigmatic possibilities?185
This book develops the hypothesis that codification, abstraction, and diffusion have to be the constituent dimensions of an information economy: the E-space indicates how increases in mutually reinforcing codifications and abstractions save on data processing, and the C-space highlights the economies in transmission that can be achieved by increases in codification – whether this is supported by abstraction or not. Codification, the information-preserving act of shedding physical data, is common to both spaces. The U-space then allows one to relate the utilities gained through the greater abstraction and diffusibility of created knowledge to the cognitive and communicative investments these require in both the E- and the C-spaces – i.e., it establishes both the demand and supply conditions for knowledge. The information economy, foreshadowed in the information production function of Chapter 1, now emerges almost naturally from intelligent attempts to deal with the scarcities that condition our own human existence as finite energy and information processing systems bounded in space and time.
One of the most important propositions to emerge from the preceding chapters is that information has a physical basis: it originates in attempts to economize on the energy and time consumed by data processing and on its spatial transmission. In a fundamental sense, then, information must be considered a product of economic behaviour itself, whether consciously intended or not. The urge to economize on energy and space-time appears to be present in all living organisms and may even inhere in nature as a whole.186 Learning systems, whether living or inert, are those which have the capacity to channel this urge to economize into the creation of new knowledge.
To claim a physical basis for information processes is not to push a reductionist thesis. Knowledge is not merely information, just as information is not merely data. The data field, as we have just hinted, displays a potential for self-organization that imparts to it an evolutionary dynamic. The next chapter examines this dynamic in greater detail.