Weaver is well aware of this sort of oscillation, by which an increase in meaning involves loss of information, and vice versa: «One has the vague feeling that information and meaning may prove to be something like a pair of canonically conjugate variables in quantum theory, they being subject to some joint restriction that condemns a person to the sacrifice of the one as he insists on having much of the other.» 2′
Information, Order, and Disorder
In his collection of essays titled Information Theory and Esthetic Perception, Abraham Moles has systematically applied information theory to music.22 He clearly accepts the notion that information is directly proportional to unpredictability and sharply distinct from meaning. What intrigues him most is the ambiguous message—that is, the message which is at once particularly rich in information and yet very difficult to decode.
We have already encountered this problem: the highest level of unpredictability depends on the highest level of disorder, where not only the most common meanings but every possible meaning remains essentially unorganizable. Obviously, this is the problem that confronts all music aiming at the absorption of every possible sound, the broadening of the available scale, and the intervention of chance in the process of composition. The dispute between the supporters of avantgarde music and its critics concerns precisely the greater or lesser comprehensibility of a sound event whose complexity transcends all the habits of the human ear and every system of probability.’ Insofar as we are concerned, the problem still involves a dialectic between form and openness. between free multipolarity and permanence, which inevitablycharacterizes the system of possibilities of a work of art.
According to information theory, the most difficult message to communicate is the one that, relying on a wider range of sensibility on the part of the receiver, will avail itself of a larger channel, more likely to allow the passage of numerous elements without filtering them—that is, a channel capable of conveying a great deal of information but with the risk of limited intelligibility. When, in his Philosophy of Composition, Edgar Allan Poe defines a good poem as one that can be read at one sitting (so as not to ruin its intended effect with interruptions and postponements), he is in fact considering the reader’s capacity to receive and assimilate poetic information.
The question of the limits of a work of art, often broached by early aesthetics, is much less banal than it might seem, since it already reveals a certain concern with the interactive relationship between the human subject and an objective mass of stimuli organized into comprehensible effects. In Moles’s study, this question, enriched by more recent discoveries in the fields of psychology and phenomenology, becomes the question of the «difference threshold in the perception of duration.» Given a brief succession of melodic data reiterated at everincreasing velocity, there soon will be a moment when the ear, having reached saturation, ceases to perceive distinct sounds and hears an undifferentiated sonic mixture.
This measurable threshold represents an insurmountable limit, and is, in itself, further evidence of the fact that a disorder which is not specifically aimed at subjects accustomed to moving among systems of probability will not convey any information. This tendency toward disorder, characteristic of the poetics of openness, must be understood as a tendency toward controlled disorder, toward a circumscribed potential, toward a freedom that is constantly curtailed by the germ of formativity present in any form that wants to remain open to the free choice of the addressee.
The distance between a plurality of formal worlds and undifferentiated chaos, totally devoid of all possibility of aesthetic pleasure, is minimal: only a dialectics of oscillation can save the composer of an open work.
A case in point is that of the composer of electronic music who, finding the unlimited realm of sounds and noises entirely at his disposal, can suddenly be quite overwhelmed by it: he wants to offer his listener the full and complex freedom of his compositions, but cannot help referring to the editing and mixing of his material, and using abscissas to channel basic disorder into matrices of oriented potential. In the end, as Moles points out, there is no real difference between noise and signal, except in intent.
Similarly, in electronic music, the difference between noise and sound is resolved by the voluntary act of the creator, who offers his audience a medley of sounds to interpret. But if he aims at both maximum disorder and maximum information, he will have to sacrifice some of his freedom and introduce a few modules of order into his work, which will help his listeners find their way through noise that they will automatically interpret as a signal because they know it has been chosen and, to some extent, organized.24
Like Weaver, Moles believes he can recognize a system of indeterminacy whereby information decreases as intelligibility increases. But he goes further: considering indeterminacy as a constant in the natural world, he expresses it with a formula that reminds him of the one used to express uncertainty in quantum physics. For if the methodology and logic of indeterminacy, borrowed from scientific disciplines, are cultural phenomena that may affect the formulation of poetics without being able to explain it, this second kind of indeterminacy, based on the correlation between freedom and intelligibility, can no longer be considered as a more or less distant influence of science on art, but should rather be seen as the specific condition of a productive dialectics—or, to use Apollinaire’s expression, the constant struggle «de l’ordre et de l’aventure,» the only condition by which a poetics of openness can also be a poetics of art.
Postscript
All these points need further clarification. It would indeed be possible to show that the mathematical concept of information cannot be applied to the poetic message, or to any other message, because information (qua entropy and coexistence of all possibilities) is a characteristic of the source of messages: the moment this initial equiprobability is filtered, there is selection and therefore order, and therefore meaning.
This objection is perfectly correct if we consider information theory only as a complex of mathematical rules used to measure the transmission of bits from a source to a receiver. But the moment the transmission concerns information among human beings, information theory becomes a theory of communication, and we need to establish whether concepts borrowed from a technique used to quantify information (that is, a technique concerned with the physical exchange of signals considered independently from the meanings they convey) can be applied to human communication.
A source of information is always a locus of high entropy and absolute availability. The transmission of a message implies the selection of some information and its organization into a signifying complex. At this point, if the receiver of the information is a machine (programmed to translate the signals it receives into messages that can be rigorously referred back to a particular code, according to which every signal signifies one and only one thing), either the message has a univocal meaning or it is automatically identified withnoise.
Things are, of course, quite different in a transmission of messages between people, where every given signal, far from referring univocally to a precise code, is charged with connotations that make it resound like an echo chamber. In this case, a simple referential code according to which every given signifier corresponds to a particular signified is no longer sufficient. Far from it, for, as we have already seen, the author of a message with aesthetic aspirations will intentionally structure it in as ambiguous a fashion as possible precisely in order to violate that system of laws and determinations which makes up the code.
We then confront a message that deliberately violates or, at least, questions the very system, the very order—order as system of probability—to which it refers. In other
words, the ambiguity of the aesthetic message is the result of the deliberate «disordering» of the code, that is, of the order that, via selection and association, had been imposed on the entropic disorder characteristic of all sources of information.
Consequently, the receiver of such a message, unlike its mechanical counterpart that has been programmed to transform a sequence of signals into messages, can no longer be considered as the final stage of a process of communication. Rather, he should be seen as the first step of a new chain of communication, since the message he has received is in itself another source of possible information, albeit a source of information that is yet to be filtered, interpreted, out of an initial disorder—not absolute disorder but nonetheless disorder in relation to the order that has preceded it. As a new source of information, the aesthetic message possesses all the characteristics proper to the source of a normal informative chain.
Of course, all this quite expands the general notion of information; but the important thing here is less the analogy between two different situations than the fact that they share the same procedural structure. A message, at the outset, is a disorder whose latent meanings must be filtered before they can be organized into a new message—that is, before they can become not a work to be interpreted but an interpreted work (for example, Hamlet is a source of possible interpretations whereas Ernest Jones’s reading of Hamlet, or T. S. Eliot’s for that matter, is an interpreted message that has condensed a disordered quantity of information into an arrangement of selected meanings).
Obviously, neither this filtered information nor the informative capacity of the sourcemessage can be precisely quantified. And this is where and why information theory becomes a theory of communication: it preserves a basic
