The super-informed society

It is at last clear that because of resource and pollution constraints, material progress has ceased to be a realistic goal for humanity.

Since the idea of 'progress' under-lies that whole shaky edifice of beliefs with which we have been so deeply imbued since our earliest childhood and which we can refer to as the world-view or Paradigm of Industrialism, psychologically, we are utterly committed to it.

We thus have no alternative but to redefine 'progress' which means determining some way in which we can 'advance' other than by accumulating material goods.

Since the quasi-religious world-view of industrialism accentuates the quantitative at the expense of the qualitative and largely ignores the notion and implications of organisation, 'progress' is conceivable by us only in terms of the accumulation of some quantity and if this cannot be of consumer goods, it must be of something else that is equally quantifiable and hence, equally easily accumulated.

'Information' admirably fits the bill. It is in the field of information processing and communications that our scientists and technologists are making the most rapid progress and in which the most spectacular advances are yet to be expected.

What is more, the energy and resources required for constructing the equipment required for processing, emitting and receiving information is supposed to be so modest and its cost so low, that its commercialisation on a massive scale is not seen as constrained by resource shortages, pollution emission standards nor the present, it is hoped, short-lived decline in purchasing power.

In these conditions, it is not difficult to persuade ourselves that the growth of the 'information' industry is really serving some useful purpose - indeed that it may well provide the next stage in the progress of man in his quest for Paradise. As Hald, one of the high priests of this new Progress, tells us:

"We are moving from a society perceived as resource-constrained to one that is 'information-rich'. We are entering a new era in which economic growth is derived from the exchange of information and the creation of knowledge rather than from the accelerated consumption of natural resources." [1]

Hald, and he seems to be echoing a view that is very widely held, believes that the proliferation of low cost computers will "fundamentally transform how we think and perceive reality":

"Individual computer capability tied into sophisticated satellite-, cable- and broadcast-based telecommunication systems, will permit millions of people to communicate simultaneously in vast interactive networks." [1]

Among other things, he contends, this will not only make people more aware but will also enable them to think properly, "relating ideas and weaving patterns of understanding, developing a form of thinking that will be highly conceptual." He goes so far as to suggest that "our children's children may become the first genius generation". According to Hald, governments will also be transformed:

"Individuals in open societies will be able to develop consensus networks through which an ongoing process of large-scale, many-to-many interaction could distil meaningful options for the future. Political leaders tied into such consensus network would, by necessity, feel closer to the voters and more committed to the directions chosen." [1]

Naive wishful thinking

That the information-rich society is technically feasible, I have no doubt. That its development might enable us to prolong for a few more decades many of the features of our moribund industrial society is also possible - though very much more doubtful. That it will create a race of supermen and make our government truly democratic, let alone solve any of our real problems, I regard as no more than the most naive wishful thinking.

To show exactly why I believe this to be so would mean covering a lot of ground. In this article, all I propose to do is to take the first logical step in this direction and consider what the term 'information' means when it is used in a precise and quantitative way. I refer to the concept of Information as developed by Shannon and Weaver. [2] I shall try to show that though their theory may be useful in the field of communications, it has little relevance to the world of living things - contrary to what is generally assumed by many scientists.

This suggests that we should reconsider exactly what 'information' is, before we can talk seriously of mass-producing it as a means of solving the problems that our society faces today.

Shannon and Weaver's concept of information

Shannon and Weaver's theory of Information was developed as long ago as 1948. Since then other theories of information have been proposed, but they seem to constitute little more than minor variations on the original theme. In any case, they do not appear to have earned any general acceptance among scientists. They are listed, together with their most salient features, by Roger and Kincaid. [3]

Both Shannon and Weaver, when they developed their theory, were working for the Bell Telephone Company. Their chief concern was to determine how to maximise the amount of 'information' - not just the number of signs - that could be transmitted via a communications channel with limited capacity. They found it convenient, for their purposes, to define information in such a way that it could be measured in terms of Boltzman's mathematical formula for the measurement of entropy. [4]

Information is thereby equated with entropy, with the difference that, whereas entropy is seen as the most probable arrangement of molecules in a particular energy state, information measures the most probable arrangement of signs in a message, both equating probability with randomness, in accordance with the Second Law of Thermodynamics, or the Entropy Law.

Randomness the "ideal"?

It is difficult to understand the philosophy underlying this notion unless one realises that, for the communications engineer, randomness (and thus the absence of any organisation or constraints on the order in which the signs appear) is equated with the freedom he enjoys in choosing the message he wishes to send and hence the order in which the signs must appear, so as to satisfy his professional requirements. Randomness, or entropy, is thus for him the 'ideal', and must thereby, for his purposes, be associated with the highest information.

To quote Weaver,

"Information is highest when the probabilities of the various choices are as nearly equal as circumstances permit - when one has as much freedom as possible when making a choice, being driven as little as possible towards some certain choices which have more than their share of probability." [2]

On the other hand, when a "situation is highly organised, it is not characterised by a large degree of randomness or of choice" and in these conditions, "the information (or the entropy) is low".

Linguistic constraints

The sort of constraints that Shannon and Weaver regard as reducing this freedom of choice (and hence the information content of a message) are linguistic constraints. Each language has a particular structure or organisation. In terms of this structure, one can predict with a measure of confidence that certain words are more or less likely to follow other words. Thus to quote Weaver,

"After the three words 'in the event' the probability for 'that' as the next word is fairly high, and for 'elephant' as the next word is very low." [2]

These linguistic constraints reduce the information content of a message by forcing the sender to include signs in his message, not because he wants to but because they are imposed on him by the structure of the language in which the message is formulated.

To him these former signs are redundant. Different languages are seen as having a different built-in redundancy, that of the English language being about 50 percent. Thus one can say that the higher the organisation and hence the lower the entropy, the greater must be the constraints, the higher must be the redundancy, and the lower must be the information contained.

Measuring information

The amount of information in a message is calculated in terms of the logarithm (base 2) of the number of choices. The result is formulated in terms of 'bits' (the term 'bit' was first suggested by John W. Tukey, as an abbreviation for 'binary digit'). When numbers are expressed in the binary system there are only two digits, zero and one. These may be taken symbolically to represent any two alternate choices. In a situation in which there are only two choices, there is said to be one bit of information.

The greater the number of free unconstrained choices, the greater the amount of information. If there are 16 choices from among which we are equally free to choose, then such a situation is associated with four bits of information.

Improbability

The greater the freedom enjoyed by the sender in the selection of signs or messages for emission, the greater must be the improbability that a particular sign or message will be sent. To illustrate this, I shall assume that Shannon and Weaver's 'information' takes 'meaning' into account. Thus, a message that told us that a horse called Green Crocodile would win a race in which there were 16 contestants of unknown breeding and with no previous form (i.e. all in theory having the same chance of winning) would communicate 4 bits of information.

If we knew something about their breeding and form, and, on this basis, could classify the horses in accordance with what appeared to be their chances of winning the race, the information communicated would be correspondingly reduced. If one horse were backed down to even money, on the theory that it had one chance out of two of winning the race, then a message informing us that it would win would communicate still less information, in fact no more than one bit - the same amount of information as it would communicate, were Green Crocodile to have but a single other contestant to deal with, rather than 15 others.

This is clearly a very sensible way of calculating the value of information from the point of view of communications. The greater the number of bits ascribed to a message, the more valuable the information must be. This is certainly so in the case cited, in any case, to both the bookmaker and the punter.

Meaning

In reality it does not quite work this way since Shannon and Weaver are not concerned with the probability or improbability of a statement being true or false. This is the concern of the epistemologist, not of the communications engineer. The latter is not even preoccupied with the probability or improbability of a particular statement, nor even of a particular word, but only of particular signs being emitted - regardless of whether these signs make up intelligible words or whether such words make up intelligible sentences.

In other words, the information content of a message, for them, does not take into account its meaning. This both Shannon and Weaver fully admit. Thus Weaver writes,

"Information must not be confused with meaning."

Likewise Shannon:

"The semantic aspects of communication are irrelevant to the engineering aspect." [2]

This means, as again they freely admit, that their use of the term 'information' is very different from its normal use in the English language.

Reducing the information content

An essential feature of Shannon and Weaver's theory is, that during the emission of a message its information content is reduced. The reason is that as a message is spelled out along a channel, so does the probability or improbability of specific signs occurring become easier to calculate. Linguistic organization is seen to build up - as does 'redundancy', which means that 'entropy' and 'information' are correspondingly reduced.

Noise

Another reason why the amount of information contained in a message must fall as it is spelled out, is that communication channels are subject to 'noise' or 'randomness'. Noise, of course, increases uncertainty or improbability. One might think that it would thereby lead to increased (rather than decreased) information. However, Shannon and Weaver distinguish between the type of uncertainty caused by noise, which they regard as undesirable and desirable types of uncertainly which they identify with "freedom of choice", and hence with information.

The information content of a message is thereby not equal to uncertainty but to 'desirable' uncertainty minus 'undesirable' uncertainty or noise.

The extension of the theory

The fact that the equations used to measure entropy and information are the same is to Weaver highly significant. He points out that for Eddington

"the law that entropy always increases - the second law of thermodynamics - holds, I think, the supreme position among the laws of Nature". [2]

Thus, Weaver notes, when the engineer

"meets the concept of entropy in communications theory, he has a right to be rather excited - a right to suspect that one has hold of something that may turn out to be basic and important." [2]

It is undoubtedly this feature of Shannon and Weaver's concept of 'information', (and by the same token its compatibility with the paradigm of science and hence with that of industrialism) which make it so attractive to the scientist and which, quite wrongly, seems to justify taking it out of its original context - that of communications engineering - and seeking to apply it to the world of living things, which of course, has only served to confuse the issue and, at the same time, to delay the development of a theory that really explained what is information and what are the principles governing its use in the world of living things.

Measurement

To begin with, one of the aspects of the Communications Theory of Information that makes it so attractive to the scientist, is its quantifiability. For quantification to be possible, however, as Apter points out, we must know the exact number of possible messages that could be transmitted at any one time. This may well be possible in the field of communications but not in the field of behaviour.

It is for this reason, as Brillouin points out, that

"the modest but, we think, significant applications of information theory to various psychological experiments have occurred in precisely those situations in which the set in question was strictly defined: a list of syllables to be memorised, associations to be formed, responses selected from etc.. There was therefore no difficulty in quantifying the associated 'amounts of information' and relating such amounts to certain aspects of performance". [5]

But, as he points out, such situations are "banal". We can add to this that they do not normally occur in the living world.

Entropy and disorder

A further consideration is that the concept of entropy itself does not apply to the world of living things, any more than does that of information. [4]

It is supposed to be equated with biospheric disorder but this means looking at the biosphere in purely energetic terms i.e. in terms of but one of its innumerable components - an error I have referred to elsewhere as energy-reductionism. [4]

An increase in entropy really means the homogenisation of temperature - and it is simplistic to equate such a process with the disintegration of a natural system. Even if we accepted that it were, and identified entropy with biospheric disorder, it is easy to show that the Entropy Law has not applied to the world of living things, which over the last 3 billion years, rather than become increasingly disorderly, has, on the contrary, moved in precisely the opposite direction - towards ever greater complexity and order. Those who still believe that the entropy law is the supreme law of the Universe try to explain away this embarrassing fact in a number of ways, but none of them are at all convincing. [4]

In reality, disorder in the biosphere and hence, if we like, entropy, rather than being a highly probable state, is, on the contrary, an extremely improbable one, as must be Shannon and Weaver's ideal source of information.

The sender and the source

In any case, the notion of a passive source of information (whether it displays order or disorder) - from which messages are selected by an external agent - does not correspond to anything that exists in the world of living things.

The natural systems that make up the biosphere are dynamic not static, active not passive and, what is more, they are self-regulating not regulated from the outside (asystemically) by an external agent such as a communications engineer.

The source of information and the sender of the message in the world of living things are, in fact, part and parcel of the same self-regulating system.

If we integrate Shannon and Weaver's sender of messages and the source of the messages into the same system, it must cease to display entropy, for one of the basic features of entropy is randomness and hence non-purposiveness, but the sender of messages acts purposefully since, as we are told, he selects for emission those messages that display the minimum redundancy, and hence the maximum information content.

What is more, if the system is to achieve its goal efficiently, then the information it contains must be organised in that way which most favours the achievement of that goal. This we can predict with confidence on the basis of our empirical knowledge of the way patterns of information (brains, genes, genomes, gene-pools etc) are organised in the world of living things.

Improbability in the biosphere

Another consideration is that the sort of improbability that Shannon and Weaver write about is not a useful concept for understanding the working of the biosphere.

For Shannon and Weaver, improbability is either improbability vis-à-vis the workings of the entropy law, which we have seen, does not apply to the world of living things, or else it is improbability on the basis of probability theory, which they wrongly take to be the same thing.

In the world of living things, improbability, if we are to use this concept, means improbability vis-à-vis a system's model or image of its relationship with its environment, which reflects its experience and that of its cultural group (if a human animal) and of its species for a specific purpose - that of assuring the system's stability vis-à-vis its environment, and hence its survival.

Thus, as living things evolve, they develop the capacity to discriminate between an increasing range of different environment situations, to interpret them correctly and to react to them adaptively. A very simple organism such as the Dyonea Fly-trap that so fascinated Darwin can, when something lands in its trap, do one of two things: close it or not and it does so with the minimum powers of discrimination, since it cannot discriminate between an edible insect and an inedible pebble.

At the other end of the scale is a human animal that can handle a vast number of different signals and interpret them correctly and thereby has at its disposal an exceptionally large repertoire of adaptive responses. In the language of Shannon and Weaver, one can say that the human animal is capable of handling messages with a high degree of improbability and hence of high information value - thousands of bits of information - as contrasted with the mere one bit that the Dyonea Flytrap can handle.

The human animal, however, cannot handle each of these messages with equal ease. Nature is incredibly efficient. The ease with which a living thing can handle messages seems to be a function of their importance or relevance to its behaviour pattern (I shall consider this question in detail further on) and also of the probability, in terms of its own experience and that of its species, that such a message will actually be received.

In other words, information in the brain and nervous system is not arranged at random and hence does not display entropy but is on the contrary highly organised - as is all information made use of by natural systems within the biosphere (genetic information for instance).

This must be so, too, since the behaviour mediated by such information is characterised by its orderliness. The moves that make up a behavioural strategy - such as the development of an embryo in the womb, the bringing up of a child within the family unit, or the cultivation of a garden in a traditional society of horticulturalists, for instance - are not arranged at random. They are highly organised.

Diversity

If information is organised, partly at least, in accordance with the probability of its being required, then systems living in a protected environment in which only probable things occur, will need to react adaptively only to a limited range of different environmental situations.

Those that live in a less well protected environment in which improbable things occur, will need to react adaptively to a wider range of more improbable environmental situations - and hence will have to make use of a correspondingly more sophisticated organisation of information, some of which may never have to be used.

Such a system is said to display behavioural diversity and to make use of information whose organization can be regarded as displaying a corresponding degree of diversity.

The behaviour that such diversity permits is referred to by Julian Huxley as "cladogenesis". Holling refers to a system capable of such behaviour as displaying "resilience". [6]

The diversity displayed by a natural system can also be regarded as its 'redundancy', or at least its apparent redundancy. It is not a measure of what a natural system does, at least in the short-term, but of what it is capable of doing. It measures all the signals it is capable of handling and all the responses it is capable of mediating, even though during the course of the system's lifetime it may never have the occasion to exploit more than a minute fraction of these possibilities.

Redundancy in natural systems

It is astonishing just how much redundancy, in this sense of the term, is built into natural systems. To give an idea; if one of our lungs is destroyed, not only can we survive without it, but we can actually tolerate the destruction of most of the second lung as well. In fact, so long as 6 percent of one lung remains, we can go on living a fairly normal existence.

However, we would no longer be capable of undue exertions. Thus we would not react adaptively to a message which told us to sprint 100 yards at breakneck speed in order to avoid being eaten by a tiger. However such a message, it may he argued, is unlikely to be received.

Lashley [6a] has also shown that we can do very well with but a small part of our neocortex. A genepool and the population it gives rise to, can also be decimated without bringing about the extinction of the species in question. When a human male ejaculates, he frees something like three hundred million spermatozoa, of these no more than one is required to fertilise a female.

Shannon and Weaver rightly regard a certain amount of redundancy as useful for counteracting the effects of noise. However they take it to be otherwise undesirable in that it reduces the information content of a message by reducing the freedom of choice of the sender and hence the variety of messages he can send.

But in the world of living things, as already mentioned, redundancy should, on the contrary, be identified with the diversity of variety of the messages that can be sent or received; rather than reduce a message's information content, it must, on he contrary increase it - since it permits the mediation of an essential aspect of behaviour, its ability to adapt to improbable events.

Information is more than improbability

An even more serious criticism of the extension of Shannon and Weaver's concept of Information is that it only provides a measure of the improbability of a message (whether it be the right or the wrong sort of improbability). Information, in the world of living things, as I have already intimated is very much more than this. This is also the view of Donald Mackay:

"To dress improbability up as a definition of information as some exponents do, seems the most unfortunate obscurantism. Unexpectedness is a measurable quality or attribute of information - not a definition of it." [7]

This is also the view of Brillouin [5]:

"it is naive to take simply the flux of signals per second, to multiply it by bits per signal in the communication engineering sense, and call the result 'amount of information' in the sense of transmission of knowledge (labelling everything one does not like, 'noise')."

First of all, in the world of living things, a message is not emitted because it is improbable, or, for that matter probable. It is emitted because it is of some relevance to the relationship between the sender and the receiver. Yet Shannon and Weaver are not in the least bit concerned with whether the receiver is interested in receiving a message, let alone whether he can understand it or is likely to believe it - all of which considerations must be of critical importance. Again this may make sense in the world of communications engineering, but not in the world of living things.

Information and its receiver

As Waddington [8] points out, information in the real world largely consists of instructions or programmes or 'algorithms'. Thus, genes combine to provide instructions for protein-synthesis. A genepool provides instructions for the renewal of a viable population. The brain and central nervous system provide instructions for the proper functioning of an individual's metabolism and, for his day-to-day adaptive relation-ships with his environment (neurogeny). A culture provides instructions for the mediation of a society's adaptive behaviour pattern.

These instructions, and hence this information, are not designed to be transmitted into a random environment. Information, as Brillouin notes, is not something "that can be poured into an empty vessel like fluid or even energy". [5] This is one of the most important things wrong with the neo-Darwinist theory of natural selection, in which behaviour is seen as determined by the genes acting in what is taken, implicitly at least, to be a random environment. [9] [10]

The genes are not dictators, as Weiss [11] puts it, but "interact in co-operation with the whole of which they are part". The instructions they issue will only be obeyed by systems that have been programmed by their evolution and upbringing to receive, understand and believe them. This must be true of the transmission of instructions and hence of information in all living processes.

As Waddington writes,

"No transmission system can effectively carry information between a transmitter and a recipient unless the recipient accepts the message as meaningful ... As the new born infant develops, for instance, it must be 'moulded' into an information accepter ... and an entertainer of beliefs...[and] unless this happens the mechanism of information transfer cannot operate." [12]

But this is not enough. The receiver of a message must also be structured in such a way as to be capable of acting on the information adaptively. As Waddington writes,

"It is no use pushing the DNA of your sperm into an egg unless the egg contains the polymerases capable of transcribing it into a messenger and all the rest of the machinery for turning out a protein according to specification."

The cries of a baby in distress provide an important message to its mother who is not only geared to hearing them and understanding their significance but also to responding to them effectively. Otherwise there would be no advantage to be gained from the ability to detect them.

Importance

The quality of a message that will determine whether it will be detected, and interpreted by a natural system is its relevance to its behaviour pattern, or (what is the same thing) its importance to it.

Since information, in a natural system, is organised hierarchically - from the general to the particular - the importance of a message can be determined in accordance with its relevance to the most general and important information contained within a pattern of information or cybernism, which in turn should reflect its relevance to the most important or general phases of the associated system's behavioural strategy.

Simple forms of life, it can be shown, are only capable of responding to messages which, in the psychological literature, are referred to as 'stimuli'. Oatley defines a stimulus as

"that aspect of an event of biological importance to a particular animal to which it is sensitive, and by which the response is controlled." [13]

This is clearly illustrated, he points out, by the behaviour pattern of the tick as described by Von Uexhull.

"In the tick's world just three events are important: each is detected in terms of the presence of a single aspect of the situation, and each stimulus to which the tick is sensitive triggers a particular response. Thus when butyric acid is detected in the air, the tick releases its grasp on the branch from which it was hanging. It happens that butyric acid is a chemical secreted by the skin of mammals, and by letting go the branch when it detects this chemical it stands a good chance of landing on the back of a suitable host passing beneath it. Just as the zoologist classifies mammals by whether they suckle their young or have fur, the tick classifies them by whether or not they produce butyric acid. But unlike the zoologist, the tick when it is in the tree is quite insensitive to any other aspect of mammals. The tick is also equipped to detect mechanical stimulation from its host's hair. This stimulus causes the response of crawling about. Lastly it detects heat, and this causes it to bore into the host's skin. Thus events relevant to the life of the tick might plausibly be detected simply by receptors sensitive to butyric acid, mechanical stimulation, and warmth." [14]

Such receptors will be capable of picking up all messages relevant or important to the behaviour pattern of the tick, the tick simply not being equipped to pick up messages of lesser importance.

The principle involved is clear to someone running a business enterprise. Among other things he must develop the ability to distinguish between the messages he receives that are of importance to his business and those that are not. Since he is likely to have few assistants and must fulfil by himself all the tasks required to assure the survival of his enterprise, he does not have the time to deal with relatively unimportant messages which he must simply ignore.

As natural systems evolve, they develop the capacity to deal with messages of lesser importance as well. This enables them to develop a correspondingly more subtle behaviour pattern which permits them to adapt with greater perfection to their specific environment. Nevertheless it will still be the more important messages with which they are primarily concerned.

To return to the analogy with the managing director of a business enterprise, we would then regard his organisation as having expanded. This means that when messages arrive which are not sufficiently important for him to take the time off to read and act upon, then, rather than simply reject them, he can now delegate them to subordinates at the appropriate 'echelon of command'.

It is partly at least because our politicians and their scientific and economic advisers have, on the whole, failed to identify the important problems faced by the societies they have been elected to govern (population growth, social breakdown, deforestation, soil-erosion, desertification, pollution etc.) and devoted their time instead to dealing with short-term 'economic' triviata, that the world is in such a terrible state.

Nature, on the other hand, has proved very much more efficient. A natural system has a built-in capacity to select messages according to their importance to its welfare and survival and act on them at the appropriate echelon of command and with the appropriate sense of urgency.

Complexity and diversity

As we have seen in the case of the tick, the simplest informational and somatic organization permits adaptation to important events. On the other hand, that type of informational and somatic organisation which must build up for a system to become capable of adapting to trivial events, I shall take to be its 'complexity'.

We thus have two types of biospheric organization: diversity, which, as it builds up, permits adaptation, to increasingly improbable events: and complexity, which, as it builds up, permits adaptation to increasingly trivial events.

Both complexity and density contribute to stability. The former by permitting ever more subtle adaptive responses to a specific environment, the latter by permitting adaptive responses (subtle or unsubtle, depending on their complexity) to many different environments.

The greater the instability of the environment, and hence the more it is likely to change, the greater must diversity develop even at the expense of complexity.

Importance and improbability

There is indeed a necessary connection between the importance and the improbability of messages, but it is not of a nature to justify Shannon and Weaver's neglect of the concept of importance and their preoccupation with that of improbability.

On the contrary, it would be more accurate to associate the importance of a message with its probability.

From the point of view of a particular species, the most important genes are those that will confer on subsequent generations the most general features of this species, those, for instance, that assure that giraffes look and behave like giraffes rather than like fiddler-crabs or dung-beetles. It is extremely probable that these genes will be present and extremely improbable that the giraffe gene-pool will give rise to populations of such alien beasties.

On the other hand, it is less important and less probable that all giraffes should display the same superficial characteristics, since as we shall see, diversity, in so far as these superficial characteristics are concerned, is the rule rather than the exception.

The same can be said for messages coming from the outside. Living things both adapt to their environment and, at the same time, modify it so that it better satisfies their requirements and thereby becomes easier to adapt to. As this occurs so there is a corresponding increase in the probability of the emission and reception of important messages indicating the presence of those environmental constituents (the presence of food, shelter or the requisite members of the family and community) whose co-operation is required for adaptive behaviour.

At the same time, important messages indicating the presence of events that threaten the generalities of the behaviour pattern of living things (such as famines, epidemics and enemy invasions) must become correspondingly improbable.

The process of adaptation can be represented graphically by reduced discontinuities or fluctuations, corresponding to the building up of increasingly stable relationships between a system and its environment.

If improbable messages are to be identified with important messages - then it can only be with the latter threatening or negative type, rather than with the former co-operative or positive variety.

In fact, the term 'important' is far from ideal, as it tends to obscure this critical distinction between messages that are important because they favour an important process, and those are important because they threaten to prevent the occurrence of this process.

The non-plasticity of general information

Stability is but another word for continuity, and if a system's behaviour is to be stable or continuous, so must the information in the light of which it is mediated.

The generalities of a system's behaviour pattern can be shown to reflect its long-term experience; the particularities, its short-term experience. [4] The former must not thereby be modifiable to satisfy short-term ends - or the system's basic continuity would be lost. They must, in other words, be non-plastic. It is easy to show that, in normal conditions, this requirement is adequately met.

Thus the genetic information formulated in the language of DNA which is transmitted from one generation to the next, and which provides the most general instructions for the reproduction of a population, is non-plastic in the short-term at least, though the position of science today is that it is non-plastic, even in the long-term.

In the same way, the basic features of a society's world-view which we associate with its basic values - are also non-plastic. People imbued with these values are not willing to compromise on them; they are taken as given or self-evident. It is because such generalities are non-plastic that the continuity of information, as we have seen, can be maintained.

Plasticity and diversity

Plasticity is of course a precondition of diversity and hence of 'cladogenesis' or 'resilience'. If information cannot be changed, there can be no alternatives to it. If, on the other hand, it can be changed very easily then the existence of alternatives makes sense.

Since general information is non-plastic, it is not surprising that it should display low diversity, nor that trivial information, on the other hand, which is highly plastic, should display such high diversity.

It may be useful to see information, as used in the world of living things, as organised into something resembling an inverted cone, which we can regard as made up of different strata like an onion.

Its generalities - chronologically the first part of the information to develop - are at the apex. They are non-plastic. There are no alternatives to them. Diversity is low or non-existent. They reflect the experience of the past and one cannot change the past. At the base are the particularities - chronologically the last to develop - the triviata that reflect the most recent experience. They are plastic. There are lots of alternatives. These strata display the highest diversity.

All biospheric organisations of information or cybernisms cannot be represented by a cone of this sort.

It would display low complexity and low diversity. It would be extremely vulnerable to change. A system equipped with such a cybernism would be unlikely to survive by itself, therefore we would be more likely to find it associated with a lot of similar systems to form a population, one whose behaviour would be characterised by fairly large oscillations.

A society possessing such an arrangement of cybernisms would display low complexity but high diversity. It would not be able to adapt with any great sophistication to its specific environment but it could survive when subjected to environments displaying a considerable degree of improbability. Alternatively we could find a system equipped with a cybernism that could be represented schematically by a very steep sided cone.

Such a cybernism would display high complexity and low diversity, which would enable a system so equipped to adapt with incredible sophistication to a highly specialised environment but not to survive were this environment to be subjected to any radical changes. Such a cybernism would be adapted to a highly protected environment such as that enjoyed by many parasites. It would be perfectly adaptive, contrary to what Holling [6] tells us, so long as it could be predicted that its environment would remain so protected.

Information increases with development

The final reason why Shannon and Weaver's theory is inapplicable to the world of living things is that the amount of information contained in a message as it is being emitted, is seen as decreasing (because of the accumulation of linguistic constraints and noise); whereas in the world of living things, the opposite is true, i.e. the information-content of a message can only increase.

Waddington [12] admits that there are a few exceptional cases in the living world in which the information-content of a message does not increase. An obvious example is the passage of electrical impulses through networks of nerves, perhaps too, the transmission of hereditary information in the chromosomes of one organism to those of its offspring. But even then, as Waddington points out,

"Biology has developed mechanisms more flexible than those used by telephone engineers". [12]

Thus a gene may mutate; when it does, the information that the offspring receives is not exactly the same as that present in its parent.

Shannon and Weaver, I suppose, would answer that a mutation is nothing more than an error in transcription and would thereby fail into the category of 'noise', which must reduce the information-content of the message rather than increase it. But this of course would not take into account the rare instances in which mutations lead to adaptive behaviour. Also there are other mechanisms such as "chromosomal deficiencies, duplications, translocations, formation of iso-chromosomes, etc., by which the amount of information can be either increased or decreased".

However, it is in the transmission of information from the genotype to the phenotype that the "limitations of the theory become of overriding importance and rapidly render it not merely useless but a dangerous snare."

Thus, the phenotype of an organism is not simply made up of all the proteins associated with all the genes present in the genotype, it is very much more than this. In Waddington's words, it is

"a highly heterogeneous assemblage of parts, in each of which there are some, but not all, of the proteins for which the genes could act as patterns, and in each of which there are also many other substances and structures over and above the primary proteins corresponding to particular genes."

It is fairly evident, as Waddington points out, that an adult rabbit running around a field contains a very much greater "amount of variety" or information than a newly fertilised rabbit's egg. How then, Waddington asks, can one deal with such a situation "in terms of an information theory whose basic tenet is that information cannot be gained?"

Waddington here seems to be associating information with the number of different things a system can do - its 'variety' or diversity, i.e. the improbability of a situation to which it can react adaptively. But the organisation of information required to mediate more complex behaviour must also build up with development.

The information content of a natural system

That the information content of a natural system increases as it becomes more complex seems clear to a number of writers, who have sought to measure a system's complexity in terms of its information-content, using Shannon and Weaver's concept of information.

Dancoff and Quastler [15] tried to do just this. They postulated that the larger the number of different components in a system, and hence the greater its complexity, the greater must be the amount of information it contains, since the higher must be the improbability of building up such a system by assembling its components in a random manner.

Unfortunately, what Dancoff and Quastler actually measured has strictly nothing in common with the sort of complexity encountered in the biosphere. This cannot be measured by adding up its component parts, because it derives its essential features, above all, from the way these parts are organised.

Biospheric organisation, Dancoff and Quastler cannot, of course, take into account, for organisation and the constraints associated with it, as Shannon and Weaver themselves point out, are associated with reduced not increased information. Thus, unless increasing complexity is associated with reduced information and the nematode Ascaris be taken to contain more information than man, Dancoff and Quastler have to ignore the all-important organisational component of complexity.

Thus Atlan [16] expresses certain reserves as the to the validity of measuring complexity in terms of information-content, because of

"Le caractère statique et uniquement structurel de la complexité dont il s'agit, a l'exclusion d'une complexité fonctionnelle et dynamique, liée non pas à l'assemblage des elements d'une systeme mais aux interactions fonctionnelles entre ces elements."

Apter criticises Dancoff and Quastler [15] on the grounds that they are only concerned with "the specification of parts with no reference to their interrelationships". [17] As Apter notes, this means that

"there would be an equal amount of information in a building and a mass of rubble, in a Shakespeare sonnet and a meaningless jumble of letters, indeed, in a living, a dead and a homogenised organism, provided only that there was the same number of building stones in each case and that the relative amounts of these needed were the same and provided the instruction list was the same length in each case." [17]

In other words, they

"overlook precisely those qualities that are generally accepted as being the significant features of developing rather than simply growing systems."

Significantly, Dancoff and Quastler themselves admit that their work yields but "crude approximations and vague hypotheses" and that their estimates are "extremely coarse". Nevertheless they insist that this is "better than no estimates at all". I do not think this is so. Mathematical calculations based on false premises and making use of inappropriate concepts can only, by virtue of the impression of great scientific accuracy that they convey, serve to mislead people and to obscure the real issues at stake.

The attitude of critical scientists

I have tried to show that the use of the communications concept of information for understanding behaviour in the world of living things cannot conceivably be justified on either theoretical or empirical grounds.

This is not altogether surprising, since it was not designed for this purpose, any more, for that matter, than was the associated concept of entropy.

This is Waddington's view too. Information theory, he points out,

"was developed in connection with a particular type of process and has limitations which make it extremely difficult if not impossible to use in many of the biological contexts to which people have been tempted to apply it." [12]

Apter makes much the same point:

"Information theory based on statistical considerations" he writes "is concerned with how data are transmitted, ignoring, however, any human factors involved." [17]

Both Atlan [16] and Brillouin [5] as we have seen, also criticise the extension of this theory to the study of the world of living things. Yet in spite of these criticisms, all these writers, with the exception of Apter still explicitly justify its use for this purpose.

Waddington [12] for instance, argues that it allows the concept "to be clearly expressed", though what I think he really means is 'quantified'. But what, one might ask, is there to be gained by quantifying a concept that corresponds to nothing in the world of living things to which it is supposed to apply? It can only serve to give an air of spurious precision to, what is in effect, little more than a fiction.

Atlan [16] also regards Shannon and Weaver's concept of information as "a valuable quantitative tool". Though he admits that information in the biosphere may be something very different, he still considers that:

"La metaphore n'est pas complètement fausse. En effet, il existe bien des cas en biologie moleculaire, assez isolés mais importants, de transmission d'information au sens rigoureux de Shannon."

This seems to be a very unconvincing argument. Indeed, that it should suffice for a theory not to be 'completely false' for it to be accepted as part of the Corpus of Science is difficult to reconcile with Science's much vaunted objectivity and accuracy.

Brillouin's [5] argument for the extension of Shannon and Weaver's theory is that if it is "to break out" of "its original habitat of bandwidths and modulations", then a proper beginning must be made, "which usually means a modest beginning" which presumably he regards Shannon and Weaver's theory as providing.

But why not allow the concept of information to remain "in its original habitat of bandwidths and modulations?" What evidence does Brillouin or anybody else provide to suggest that its use can profitably be extended to other fields for which it was not designed? The answer, I am afraid, is none whatsoever.

On the contrary, the only function that the extension of the theory is likely to serve is to perpetuate the myth that behaviour is atomised, and random, since the theory attributes precisely such features to the information in the light of which behaviour is mediated. This can clearly only serve to obscure important features of the behaviour of living things, such as, its goal-directedness, its stability and its organisation.

This brings us to the real reason why many of our scientists have accepted the extension of the use of Shannon and Weaver's theory to so many other fields.

Scientists, and in particular, aristo-scientists, are committed to that view, of the world that we can refer to as the 'paradigm of science' - the only one that justifies the performance of those tasks that they have been trained to perform, and on whose performance hinges their status as the high priests of our industrial society.

In terms of the paradigm of science, behaviour must, above all, be seen as atomised and random, i.e. as disorganised and goal-less. Otherwise, how can they justify induction - the random accumulation of data - as the basic method of acquiring knowledge?

How else can they justify the 'analytic' or 'reductionist' method which consists in breaking things up into their component parts, and hence in systematically eliminating, as a prelude to their scientific study, whatever organisation they might have previously displayed?

How else can they justify examining systems in controlled laboratory conditions and hence in isolation from all the other systems with which they have co-evolved, and in the context of which their true goal-directed function can only be determined?

How else can they justify quantification - that sine-qua-non of scientific method - unless the accent is on measurable components rather than on their unmeasurable organisation?

Finally, how else can they justify 'statistical method', whose basic postulate, as Needham [18] tells us, is that the laws of the biosphere are but words we give to statistical regularities? This being so, it is not difficult to see the attraction to scientists of Shannon and Weaver's theory.

By defining information the way they have, they have done the scientific world a truly great service. They have contributed to the coherence of that most unsatisfactory corpus of knowledge that we call science and enable it to embrace that much more of the knowledge that could, if otherwise organised, help us to understand, which at present we do not, the essential features of the world we live in and what we are doing to it.

They have also done our economists and industrialists a good turn. Information, that is both random and atomised, whose value is neither dependent on its meaning, its accuracy or its relevance, and that is measured in terms of anonymous 'bits', provides those who have mastered the technology of computers and microelectronics, with the ideal commodity for mass-production, mass-commercialisation and mass-accumulation.

They have also provided them all with the theory in terms of which it is possible to rationalise and hence legitimise, in the most 'scientific' and hence the most credible language possible, the blind and euphoric hope that such a technology may be creating for us a new paradise on earth - the Information Rich Society - one, that in the light of the latest scientific breakthroughs, may appear less speculative than the other now largely discarded paradises of our disillusioned past.

References