Chapter 2 Chapter 1: Cybernetics in History

I have been working in many branches of information theory since the end of World War II.In addition to the electrotechnical theory of message passing, there is a broader field of information theory that includes the study not only of language but of messages as a means of machine and social control, including computers and other such automatons. development, including some psychological and neurological considerations and a new scientific methodology of a tentative nature.This broader information theory is a probability theory, but W.Inherent part of the trend of thought initiated by Gibbs, which I have already mentioned in the preface.

Until recently there was no ready-made word for this compound idea, and in order to sum up the whole field in one word, I felt compelled to invent a new one.Thus, there is the word "cybernetics", which I deduced from the Greek word Kubernetes or "helmsman", and the English word "governer" (manager) is the final extension of this Greek word.Later I happened to find that the word had already been used in political science by AmPere, and it was also quoted from another angle by a Polish scientist, both in the early nineteenth century. I once wrote a more or less technical book entitled "Cybernetics", which was published in 1948.In response to everyone's request to make the concept of cybernetics acceptable to the general public, I published the first edition of a book in 1950.Since then, the discipline has grown from a handful of ideas jointly developed by Drs. Claude Shannon, Dr. Warren Weaver, and myself into a definite field of study.

Therefore, I took the opportunity of re-versioning the book to rewrite it in line with the latest situation, and at the same time delete some shortcomings and inconsistencies in the structure of the original book. In the definition of cybernetics given in the first edition, I grouped communication and control together.Why do I do this, when I correspond with someone, I send him a message, and when he messages me back, he sends back a related message that contains what he understood first and not mine information.When I go to control someone's actions, I have to send him a message, and although the message is imperative, the technique of sending it is no different from the technique of reporting a fact.

Moreover, if my control is to be effective, I must hear from him any information which indicates whether the order was understood or not and whether it was carried out. The main theme of this book is to clarify that we can only understand society through the study of information and the study of social communication devices; to clarify the future development of these information and communication devices. The news between people is bound to occupy an increasingly important position in society. When I give a command to a machine, the situation is not fundamentally different from when I give a command to a man.

In other words, as far as my consciousness is concerned, all I know are the commands sent and the acknowledgments sent back.For me personally, it does not matter whether the signal passes through a machine or a person in its intermediary stage, and in any case it does not make me too attached to the signal. The change.Therefore, engineering control theory, whether human, animal or mechanical, is an integral part of information theory. Of course, there are differences of all sorts of details both in the messages and in the problems of control, not only between living beings and machines, but also within their respective smaller spheres.The purpose of cybernetics is to develop language and various technologies so that we can truly solve the general problems of control and communication, but it also needs to find a set of special ideas and technologies to distinguish control and communication under the guidance of certain concepts. various special forms of expression.

The commands we use to control our environment are all information we give to the environment.These commands, like any form of information, are broken up in transit.They generally arrive in a less legible form, certainly no more legible than when they were sent.In control and communication, we must fight against the natural tendency to reduce organization and impair meaning, what Gibbs called the tendency to increase entropy.Much of this book speaks of the limits of communication within and between individuals.Man is bound in the world that he can perceive with his own senses.Whatever information he receives has to be adjusted through his brain and nervous system, and only after a specific process of storage, collation and selection does it enter the effectors, usually his muscles.These effectors act on the outside world, and at the same time react to the central nervous system through receptors such as the end of the motor sensory organs, and the information received by the motor sensory organs is combined with the information stored in the past to affect the future. action.

The content of the name information is what we exchange with the outside world when we regulate it and make our regulation known to the outside world.The process of receiving information and using information is the process by which we adjust to various contingencies in the external environment and live effectively in this environment.The needs and complexity of modern life have placed unprecedented high demands on the information process, and our publishing houses, museums, science laboratories, universities, libraries and textbooks have to meet the various needs of the process, otherwise they will be lost. lose their purpose.To live effectively is to have enough information to live.From this it follows that communication and control are as essential to the inner life of the individual as they are to the social life of the individual.

The study of communication problems occupies a place in the history of science that is neither insignificant and accidental nor unprecedented.Problems of this kind were popular in physics long before Newton, especially in the work of Fermat, Huggens, and Leibnitz; They are interested in physics, and the focus of their interest is not mechanics, but optics, that is, the problem of the transmission of visual images. Fermat advanced the study of optics with his principle of minimization, which states that light passes in the least amount of time in any sufficiently short interval of the optical path.Huygens formulated what is now known as the "Huygens principle" in its original form, which states that when light propagates from a source, it forms something like a small sphere around the source, which Consists of secondary light sources whose light proceeds in exactly the same way as the primary light source.Leibniz, on the other hand, saw the whole world as a collection of substances called "monads" whose activity was mutual perception on the basis of a predetermined harmony arranged by God, and, quite clearly, Leibniz Nietzs considered this interaction primarily in optical terms.Monads have no "windows" other than this perception, so that, according to Leibniz, all mechanical interactions are really nothing but subtle corollaries of optical interactions.

In this aspect of Leibniz's philosophy, the author's predilection for optics and information is evident everywhere.This preference is fully expressed in his two most fundamental concepts, namely: Characteristica Universalis, or universal scientific language; and Calculus Ratiocinator, or logical calculation.Although the Calculus Ratioci-nator was not perfect at the time, it was the direct ancestor of modern mathematical logic. Leibniz, who was dominated by communication ideas, was in many ways an intellectual precursor to the ideas in this book, since he was also interested in machine computation and automata.In this book, my views are far from those of Leibniz, but the problems I discuss are purely Leibniz's.Leibniz's calculating machine was only an expression of his interest in the language of calculation, the inferential calculus, which, in his mind, was only a generalization of his idea of ​​an entirely artificial language.It follows that, even with his calculating machine, Leibniz's preference was primarily for language and communication.

By the middle of the last century, C.The work of Maxwell and his forerunner, Faraday, again drew the attention of physicists to optics; light was now regarded as a form of electricity, which in turn could be reduced to the mechanism of a medium, which was strange. The solid, hard but invisible to the naked eye is called ether.At that time, it was assumed that ether was permeated in the atmosphere, interstellar space and all transparent substances. c.Maxwell's work in optics consisted in the mathematical development of previous Faraday's convincing but not mathematically expressed ideas.Jiantai's research has raised some questions to people whose answers are very vague, for example, the problem of the movement of matter through the ether.The famous experiments of Michelson and Morley in the nineties were carried out to solve this problem, and the experiments gave a completely unexpected answer: it is absolutely impossible to prove the motion of matter through the ether.

For the various questions raised by this experiment, it was Lorentz's answer that gave a satisfactory answer for the first time.Lorentz pointed out: If we regard the forces that bind matter as themselves being electrical or optical in nature, then we should expect the opposite result from the Michelson-Morley experiment.In 1905, however, Einstein transformed these Lorenz ideas into the following form: the unobservability of absolute motion is not so much a matter of any particular structure of matter as a matter of physics. postulate.From our point of view, the important point in Einstein's work is that light and matter are on an equal footing, which is the same point made in Newton's previous writings, rather than putting everything together as Newton did. Under Matter and Mechanics. When Einstein explained his views, he used observers in various ways: observers can be both stationary and moving.In Einstein's theory of relativity, it is impossible to introduce the observer without simultaneously introducing the idea of ​​a message, without in fact reemphasizing the quasi-Leibnizian state of physics (whose tendencies are still optical).Einstein's theory of relativity and Gibbs' statistical mechanics are very different things.Einstein basically used absolutely strict dynamics terms to discuss problems, and did not introduce the concept of probability, which is the same as Newton. In contrast, Gibbs' work is probabilistic from the first step.However, the direction of work of these two people represents the replacement of the viewpoint of physics, that is, in a sense, the world that exists just at the time of observation is used to replace the world that actually exists, while the ancient naive realism in physics makes For something that might make Archbishop Barclay smile. At this point, it is appropriate for us to discuss some of the concepts related to entropy mentioned in the preface to this book.As mentioned earlier, the idea of ​​entropy represents several extremely important differences between Gibbs mechanics and Newton mechanics. In Gibbs' view we have a physical quantity which does not belong to our external world, but to a set of possible external worlds, and thus appears in the set of certain questions we can ask about this external world. in the answer.Instead of exploring the outer universe, which can be regarded as a comprehensive answer to all relevant questions, physics now becomes a bill for answers to some extremely limited questions.In fact, what we are studying now has nothing to do with all possible input and output messages that we can receive and send out. What we are concerned with is only the very special input and output message theory, including such messages only given to Our finite measure of information is included. Enrollment itself is a form of pattern and organization.Indeed, we can regard collections of messages as having entropy in them, just as we do collections of external world states.Just as entropy is a measure of disorganization, the information contained in a collection of messages is a measure of the organization of that collection.In fact, the information that a message has is essentially visible interpreted as the negentropy of the message, interpreted as the negative logarithm of the probability of the message.That is to say, the more visible the message, the less informative it is.A cliché, for example, has less meaning than a great poem. I have already mentioned Leibniz's interest in automata which happened to be shared by his contemporary Pascal, who had a real contribution to the development of what we now call the table-top adding machine. contribution.Leibniz regarded the uniformity of time given by clocks set at the same instant as a model of monadic predetermined harmony.This is because the automaton technology that embodied his time was that of the watchmaker.Let us examine the movements of the little figure dancing on the top of the music box.They move according to a pattern, but this pattern is pre-arranged, and the past activities of the little people are actually irrelevant to the pattern of future activities.The chance of them deviating from the intended pattern is equal to zero.It is true that there is also a transmission of messages here, but only from the mechanism of the music box to the little man, and there it stops.Except for the above-mentioned one-way communication with the music box's predetermined harmony mechanism, the little man has no trace of communication with the outside world.They are blind and deaf and dumb things.It is inseparable from the fact that the agreed mode changes its activities. Compare their behavior with that of man, or that of any animal of moderate intelligence, such as a kitten.I called the kitten and it looked up at me.I send it a message and it receives it with its senses, as can be seen in its actions.The kitten was hungry, so it screamed.At this point it is the sender of the message.When the kitten is playing with a small hanging ball of string, when the ball swings to the left, the kitten will grab it with its left paw.At this time, within the kitten's own nervous system, through certain nerve endings such as its joints, muscles, and bones, messages of a very complex nature are sent and received; with the help of nerve messages sent by these organs, the animal can perceive to the actual location of one's own tissue and its tension.It is only through these organs that the East and the West are possible, such as the manual skill of man. I have compared the pre-arranged behavior of the little people on the music box on the one hand with the contingent behavior of men and animals on the other.We must not, however, conceive of the music box as a model of all mechanical behaviour. The older machines, and especially the older attempts at automatons, were in fact based on closed clocks.However, modern automatic machines, such as self-controlled missiles, near-explosion fuses, automatic door opening devices, control instruments in chemical plants, and other modern automatic machine equipment that perform military or industrial functions, all have sensory organs, that is, they have the ability to receive external information. receiver.They can be as simple as photoelectric cells, which undergo electrical changes when light falls on them, allowing light to be recognized in dark places, or as complex as a television set. They can measure tension from the change in conductivity produced by a wire subjected to tension, or by means of a thermocouple (this instrument is composed of two different metal sheets in contact with each other, when one of the contact points Electric current is generated when heating) to measure temperature.In the scientific instrument maker's treasure trove, each instrument is a possible sensory organ whose readings can be recorded from a distance through the intervention of specialized electronics.It follows from this that a machine of this kind is conditioned by its relation to the outside world, and thus by what happens outside.We have this machine now and have had it since sometime. Machines acting on the outside world by means of messages are also common.Photoelectric door openers are something everyone who passes through Penn Station in New York knows, and they are used in many other buildings as well.When a message was sent to the instrument that a ray of light was cut off, the message stimulated the door and caused it to open, allowing the traveler to pass. From a machine of this type that stimulates the senses to a machine that performs a task, there are many steps, either as simple as in the case of electric switches, or of virtually any degree of complexity within the limits of our engineering skills. machine.A complex action is one in which data, which can contain a large number of combinations (we call it inputs), is introduced in order to obtain an effect on the outside world (we call it an output).These combinations include both the data put in at the moment and the data stored in the past (we call it the record taken out of the memory device. These combinations are all recorded in the machine.The most complex machine ever built, capable of converting input data into output data, is the fast electronic computer, of which I intend to speak in some detail later.The behavior of these machines is determined by specific inputs, often formed of punched cards, punched paper tape, or magnetized wire, which determine how the machines operate in a different way than they have done in the past.In control, since punched tapes or magnetic tapes are often used, the data put into these machines to instruct the operation of the machines to combine information are collectively called program tapes. I have said that both humans and animals have a sense of movement, and they use this to keep records of the position and tension of their muscles.In order for any machine to take effective actions in response to a changing external environment, it must be provided with information about the consequences of its own actions as an integral part of the information needed to keep it moving.For example, when we operate an elevator, it is not enough to open the gate of the elevator bar, because the order we give should cause the elevator to reach the door just when we open the gate.The important point is that the release mechanism for opening the door is determined by the fact that the elevator actually reaches the door, otherwise, if there is nothing blocking the elevator, the passengers will step into the empty well.This kind of control based on the machine's actual performance (actual Performance) rather than its expected performance (expected Performance) is feedback; the machine needs to use various sensory elements for this control, and these sensory elements are stimulated by the activation element , they perform the function of predictors or monitors, that is, of giving instructions on a performance.It is the function of these institutions that brings under control the dynamic tendency of the organization to disintegrate, that is, they bring about a temporary and partial reversal of the normal direction of entropy. I just mentioned elevators as an example of feedback.There are many other examples where the importance of feedback is even more pronounced. For example, the cannon sighter takes information from his sight and transmits it to the cannon, which is aimed in a certain direction and causes the shell to hit the live target at a certain moment.But artillery is meant to be used in all weather conditions.Under certain climatic conditions, the lubricating oil warmed up, and the cannon turned very quickly.In other climates the oil was frozen or mixed with sand, and the cannons were slower to answer our commands.When the cannon responds poorly to our orders, and lags behind them, the error of the sighting hand is reduced if we reinforce them by giving it a supplementary advance.Usually, in order to achieve as accurate a performance as possible, we add a feedback control element to the cannon to record the degree to which the cannon lags behind the specified position, and then use this difference to give the cannon a supplementary boost. Indeed, precautions must be taken to prevent this advance from being too violent, which would cause the cannon to overshoot its designated position, and to bring it back would have to be brought back by a series of oscillations which might become more and more severe. large, which leads to serious instability.If the feedback system itself is controllable, in other words, if its own entropy tendencies can be contained by other control mechanisms and kept within sufficiently strict limits, then the above situation will not happen, and the existence of feedback is Added stability to cannon performance.In other words, the performance of the cannon has little to do with the friction load, or it can be said that the performance of the cannon is not delayed due to the adhesion of the lubricating oil. There is something very similar to the above in human activity.When I go to pick up a cigar, I don't intend to use certain muscles.In many cases, I really don't know which muscles they are.All I did was bring into play a certain feedback mechanism, a certain reflex, in which the effect of the cigar I hadn't yet achieved became a new, strengthened engagement with a lagging muscle, whatever it was.According to this method, a completely identical voluntary meeting will enable us to start from various initial positions to perform the same task, regardless of the reduction of stretching capacity due to muscle fatigue.Likewise, when I drive a car, the set of controls I have over the vehicle is not solely dependent on my impressions of the road and the driving tasks I have to do with it.If I notice that the vehicle is too far to the right of the road, that discovery will cause me to steer it to the left.This kind of control is dependent on the actual performance of the vehicle, not just on the road conditions; it's this kind of control that allows me to drive a light Austin or drive a heavy truck with roughly the same efficiency. , There is no need to form different driving habits in order to drive the two. I will have more to say about this in the chapter of this book devoted to machines, where we will argue that the study of defects in performance-defective, human-like mechanisms of machines can be useful for neuropathy. contribute to science. My contention is that the physiological activity of living individuals and the operation of certain newer types of communication machinery are quite comparable in their similar attempts to control entropy through feedback.They both have sensory receptors as part of their cycle of operation: that is, both use low-level special instruments to gather information about the outside world and use it in their operations. In both cases, the external information is not obtained neatly, but through the transforming power within the instrument, whether the instrument is alive or dead.The information is then transformed into a new form that can be used in subsequent stages of performance.This performance is valid externally in animals as well as in machines.In animals and machines, what is reported to the central regulator is not only the activities they intend to do with respect to the outside world, but also the activities they have performed with respect to the outside world.This complexity of behavior is not understood in general and does not play its due role in our everyday analysis of society; Organic reactions can also be understood accordingly.I do not mean that sociologists do not recognize the existence and complexity of social communication, but sociologists have until recently tended to ignore social communication as the glue that holds the building of society together concrete. In this chapter we present a complex set of ideas that, until recently, have not been adequately linked.These are: the view of contingency in physics introduced by Gibbs as a modification of the traditional Newtonian convention; Augustine's claim to order and to our behavior in terms of this contingency; The theory of communication between society and society as a sequence of temporal events, which, while inherently contingency, always seeks to check the natural tendency to disorder by regulating its constituent parts for various purposes.It now appears that these concepts are basically unified.