First published by Simon & Schuster, 1985. Copyright Howard Rheingold, 1985. This book is out of print; all rights have reverted to the author. Feel free.
Today, when molecular biologists talk about the "coding" of the DNA molecule, cognitive scientists discuss the "software of the brain," and behavioral psychologists write about "reprogramming old habits," they are all making use of a scientific metaphor that emerged from the technology of computation, but which has come to encompass much more than the mechanics of calculating devices. Cybernetics, the study of communication and control in physical and biological systems, was born when yet another unusual mind was drawn into the software quest through the circumstances of war.
Because of the discoveries of Norbert Wiener and his colleagues, discoveries that were precipitated by the wartime need for a specific kind of calculating engine, software has come to mean much more than the instructions that enable a digital computer to accomplish different tasks. From the secrets of life to the ultimate fate of the universe, the principles of communication and control have been successfully been applied to the most important scientific puzzles of our age. These principles were discovered through a strange concatenation of events, and the people who were involved in those events were no less unusual than the software patriarchs who preceded them.
Eccentrics and prodigies of both the blissful and agonized varieties dominated the early history of computation. Ada Lovelace, George Boole, John von Neumann, Alan Turing, and Presper Eckert were all in their early twenties or younger when they did their most important work. All except Eckert were also more than a little bizarre. But for raw prodigy combined with sheer imaginative eccentricity, Norbert Wiener, helmsman of the cybernetic movement, stands out even in this not-so-ordinary crowd.
Norbert's father, a Harvard who was a colorful character in his own right, had definite opinions about education, and publicly declared his intention to mold his young son's mind. Norbert was to become a lovingly but systematically engineered genius. In 1911, an article in a national magazine reported these plans:[1]
Professor Leo Wiener of Harvard University . . . believes that the secret of precocious mental development lies in early training . . . He is the father of four children, ranging in age from four to sixteen; and he has the courage of his convictions in making them the subject of an educational experiment. The results have . . . been astounding, more especially in the case of his oldest son, Norbert.
This lad, at eleven, entered Tufts College, form which he graduated in 1909, when he was only fourteen years old. He then entered Harvard Graduate School.
Norbert completed his examinations and his doctoral dissertation in mathematical logic when he was eighteen, then studied with Bretrand Russell in Cambridge and David Hilbert in Göttingen, where he later crossed paths with von Neumann, nine years his junior, also a student of Hilbert's, and a world renowned authority in several of Wiener's fields of interest. One of the most immediate differences between the two prodigies, even this early in their careers, was the pronounced contrast between their personalities.
Rare was the teacher or student who failed to be charmed by von Neumann, who went out of his way to assure fellow humans that he was just as mortal as everyone else. Weiner, an insecure, far less worldly, sometimes vain, and often hypersensitive personaluity, simply didn't go to as much trouble to make an impression outside the realm of mathematics, where he was confident to the point of arrogance. Bertrand Russell wrote of Wiener, in a letter to a friend:[2]
Like Babbage, Wiener was famous for the feuds he carried on. While a student at Göttingen, he impressed the administrative head of the university, Richard Courant, but Wiener accused him of misappropriating several of the younger man's mathematical ideas and appending Courant's own name to them. When he returned to Cambridge, the outraged young genius turned his energies to a novel that was never published, about someone who bore a remarkable resemblance to Courant, and who was depicted as a man who stole the ideas of young geniuses.
Before World War I, Wiener wrote pieces for Encyclopedia Americana, taught philosophy at Harvard and mathematics at the University of Maine. During World War I, Private Wiener was assigned to the U.S. Army's Aberdeen proving Grounds in Maryland, where he was one of the mathematicians responsible for the computation of firing tables. His service in 1918 was one of the reasons it was natural for Wiener's friend Vannevar Bush to think of Norbert thirty years later, when the allies needed a way to put firing tables directly into the radar-guided mechanism of antiaircraft guns.
After the end of World War I, Norbert Wiener joined the Massachusetts Institute of Technology as an instructor of mathematics. It turned out to be the beginning of his lifelong association with that institution. By the early 1920s, like his fellow polymath across the Atlantic, Wiener was turning out world-class papers in mathematics, logic, and theoretical physics. At MIT Wiener began his long friendship with Vannevar Bush, a man who in the early 1930s was deeply involved in the problems of building mechanical calculators, and in the 1940s took charge of the largest-scale administration of applied science in history.
Decades later, Wiener quarreled with his lifelong friend because Bush didn't side strongly enough with Wiener in his feud with two other colleagues. Such feuds were one of the more well-known characteristics of Wiener's style--he tended to take disagreements over scientific issues as personal attacks, even if the disputes involved his closest personal friends. Like Babbage, his judgment did not always seem equal to his imagination.
It must be said that that Wiener did have many warm lifelong friendships that didn't go sour. For all his moodiness and paranoia, Wiener truly cared about "the human use of human beings" (as he was to title one of his later books on the implications of cybernetics), and passionately reminded the scientific community of their special responsibilities regarding the apocalyptic weaponry they had created. Despite his failure to get along with some of his colleagues, Wiener never wavered in his belief that the future of scientific enterprise lay in interdisciplinary cooperation. His friendship with the physiologist Arturo Rosenbluth, and their shared dream of stimulating such interdisciplinary pursuits, catalyzed the origins of cybernetics. But Wiener might never have worked with Rosenblueth if it wasn't for the Battle of Britain.
Like von Neumann, Wiener's most important need was for interesting problems. Like von Neumann, he knew that the quantum revolution was the most interesting problem of the 1920s. And one of the effects of quantum physics on the young mathematician's thinking was to convince him that some of the most interesting problems of purely theoretical mathematics could end up having the most concrete applications in the real world.
Another effect of quantum physics was the importance of probability and statistical measures for dealing with phemomena based on uncertain information. Wiener's familiarity with these concepts was to mature under unexpected circumstances. Like von Neumann and Goldstine and Eckert, in the late 1930s Wiener wasn't yet aware that ballistics would be the avenue for bringing his knowledge of probability and statistics to bear on the most pragmatic problems, eventually to yield most astonishing results. But, like them, he would soon come to understand that his war-related task was leading to profound scientific consequences far beyond the bounds of ballistics.
The scene was set for the emergence of Wiener's astounding results, not by any series of scientific events, but by the political circumstances of the early 1940s. When war broke out in Europe, Bush assigned Wiener to the antiaircraft control project at MIT, under the direction of Warren Weaver, himself a distinguished mathematician. It seemed like a natural step for Wiener, considering his prior experience in the early ballistic calculation efforts at Aberdeen during World War I.
The key ideas that led to computers were in the air in the late 1930s, albeit in the rather rarefied air of metamathematics and other esoteric intellectual disciplines. The necessities of was and the coordinated scientific effort that they entailed served to bring those key ideas together with the few people who were equipped to understand them more quickly and urgently than might have happened in more normal times.
Von Neumann and Goldstine's accidental meeting at Aberdeen was fortuitous and unlikely, but it could hardly be called incredible. One of the circumstances that brought Wiener together with the problem of antiaircraft guns, however, was downright weird. The technological turning point of the Battle of Britain, and a critical chapter in the science of communications systems in machines and organisms, originated when a young Bell Laboratories employee in America had an odd dream. The crucial dream was not about mathematics or engineering problems connected with computers, but was related to technical issues involving antiaircraft artillery. And it was the question of how to deal with dive bombers that was the rather urgent if indirect problem that led to Wiener's later insights.
The pathway between military strategy and scientific theory was far too circuitous, coincidental, and unlikely to have been predicted in advance, and became clearly discernible only in retrospect. In many respects, the birth of cybernetics was the kind of story more likely to be found in a novel than in a scientific journal. One of the historical coincidences was the position of Vannevar Bush as the leader of war-related research. In his role as a research administrator, Bush knew that antiaircraft technology was one of his top priorities. As a scientist, MIT researcher, and friend of Norbert Wiener's, bush was also concerned with the task of building high-speed mechanical calculators.
The allies' two most pressing problems in the early years of World War II were the devastating U-boat war in the North Atlantic and the equally devastating Luftwäffe attacks on Britain. Turing's secret solution to the naval Enigma machine was responsible, in large part, for solving the U-boat problem. But where Turing's problem was one of cryptanalysis, of mathematically retrieving the meaning from a garbled message, the Luftwäffe problem was one of predicting the future: How can you shoot at a plane that is going as fast as your bullets?
Radar made it possible to track the positions of enemy aircraft, but there was no way to translate the radar-provided information into a ballistic equation quickly enough to do any good. And attacking airplanes had a disconcerting habit of taking evasive action. Vannevar Bush was well acquainted with the calculation problem when Bell Laboratories came to him with an interesting idea for an electrically operated aiming device. That is where the young engineer's dream came in.
His name was D. B. Parkinson, and he was working with a group of Bell engineers on an automatic level recorder for making more accurate measurments of telephone trransmissions--a "control potentiometer," they called it. In the spring of 1940, Parkinson had the following dream: [3]
The electrical device, as it happened, was a good start on an automatic aiming mechanism. But very serious theoretical and mathematical problems, having to do with the way the control device sent and received instructions, cropped up when they tried to construct such a mechanism. That is when Bush turned to Weaver and Wiener.
During this wartime mathematical work related to radar-directed antiaircraft fire, Wiener recognized the fundamental relationship between two basic problems--communication and control. The communication problem in the earliest days of radar was that the radar apparatus was like a badly tuned radio receiver. The true signal of attacking planes was often drowned out by false signals--noise--from other sources. Wiener recognized that this too was a kind of cryptography problem, if the location of the enemy aircraft is seen as a message that must somehow be decoded from the surrounding noise.
The noisy radar was more than an ordinary "interesting problem," because once you understand messages and noise in terms of order and information measured against disorder and uncertainty, and apply statistics to predict future messages, it becomes clear (to a mathematician of Wiener's stature) that the issue is related to the basic processes of order and disorder in the universe. Once it is seen in statistical and mathematical terms, the communication problem leads to the heart of something more important, called information theory. But that branch of the story belongs to Clause Shannon as much as, or more than, it does to Wiener.
The control problem was where Wiener, and his very young and appropriately brilliant assistant, an engineer by the name of Julian Bigelow, happened upon the general importance of feedback loops. Assuming that it is possible to feed information about a plane's path into the aiming apparatus of a gun, how can that information be used to predict the probable location of the plane? The use of statistics and probability theory was one clue. A method for predicting the end of a message based on information about the beginning was another clue. The device in Parkinson's dream was another clue.
Then it occurred to Wiener and Bigelow that the human organism had already solved the problem they were facing. How is any human being, or a chimpanzee for that matter, able to reach out a hand and pick up a pencil? How are people able to put one foot in front of the other, fall face-forward for a short distance, and end up taking a step? Both processes involve continuous, precise readjustments of muscles (the servomechanisms that move the gun), guided by continuous visual information (radar), controlled by a continuous process of predicting trajectories. The prediction and control take place in the nervous system (the control circuits of the animating automata).
Wiener and Bigelow looked more closely at other servomechanisms, including self-steering mechanisms as simple as thermostats, and concluded that feedback is the concept that connects the way brains, automatic artillery, steam engines, autopilots, and thermostats perform their functions. In each of those systems, some small part of the past output is fed back to the central processor as present input, in order to steer future output. Information about the distance from the hand to the pencil, as seen by the eye, is fed back to the muscles controlling the hand. Similarly, the position of the gun and the position of the target as sensed by radar are fed back to the automatic aiming device.
The MIT team had wondered whether someone more informed about neurophysiology had come across analogous mathematics of pencil pushing, with similar results. As it happened, there was another team that, like Wiener and Bigelow, was made up of one infant prodigy and one slightly older genius, by the names of Pitts and McCulloch respectively, who were coming down exactly the same trail from the other direction. A convergence of ideas that was both forced and fortuitous, related to but distinctly different from the convergence on digital computation, was taking place under the pressure of war.
Even von Neumann was due to get into the act, as Wiener wanted him to do--Wiener persuaded MIT to try to outbid Princeton for von Neumann's attentions after the war. Politically, militarily, and scientifically, Wiener's corner of the plot was getting thick. The antiaircraft problem, the possible explanations for how brain cells work, the construction of digital computers, the decoding of messages from noise--all these seemingly unrelated problems were woven together when the leading characters were brought together by the war.
The founding of the interdisciplinary study that was later named cybernetics came about when Wiener and Bigelow wondered whether any processes in the human body corresponded to the problem of excessive feedback in servomechanisms. They appealed to an authority on physiology, from the Instituto Nacional de Cardología in Mexico City. Dr. Arturo Rosenblueth replied that there was exactly such a pathological condition named (meaningfully) the purpose tremor, associated with injuries to the cerebellum (a part of the brain involved with balance and muscular coordination).
Together the mathematician, the neurophysiologist, and the engineer plotted out a new model of the nervous system processes that they believed would demonstrate how purpose in embodied in the mechanism--whether that mechanism is made of metal or flesh. Wiener, never reluctant to trumpet his own victories, later noted that this conception "considerably transcended that current among neurophysiologists."[4]
Wiener, Bigelow, and Rosenblueth's model, although indirectly derived from top-secret war work, had such general and far-reaching implications that it was published under the title "Behavior, Purpose and Technology," in 1943, in the normally staid journal Philosophy of Science.[5] The model was first discussed for a small audience of specialists, however, at a private meeting held in New York in 1942, under the auspices of the Josiah Macy Foundation. At that meeting was Warren McCulloch, a neurophysiologist who had been corresponding with them about the mathematical characteristics of nerve networks.
McCulloch, a neurophysiologist based at the University of Illinois, was, naturally enough in this company, an abnormally gifted and colorful person who had a firm background in mathematics. One story that McCulloch told about himself goes back to his student days at Haverford College, a Quaker institution. A teacher asked him what he wanted to do with his obviously brilliant future:[6]
McCulloch started to work with a young logician by the name of Walter Pitts. Pamela McCorduck, a historian of artificial intelligence research, attributes to Manuel Blum, a student of McCulloch's and now a professor at the University of California, the story of Pitt's arrival on the cybernetic scene. At the age of fifteen, Walter Pitts ran away from home when his father wanted him to quit school and get a job. He arrived in Chicago, and met a man who knew a little about logic. This man, "Bert" by name, suggested that Pitts read a book by the logician Carnap, who was then teaching in Chicago. Bert turned out to be Bretrand Russell, and Pitts introduced himself to Carnap in order to point out a mistake the great logician had made in his book.
Pitts studied with Carnap, and eventually came into contact wilt McCulloch, who was interested in consulting with logicians in regard to his neurophysiological research. Pitts helped McCulloch understand how certain kinds of networks--the kinds of circuits that might be important parts of nervous systems as well as electrical devices--could embody the logical devices known as Turing machines.
McCulloch and Pitts developed a theory that regarded nerves as all-or-none, on-or-off, switchlike devices, and treated the networks as circuits that could be described mathematically and logically. Their paper, "A logical Calculus of the Ideas Immanent in Nervous Activity," was published in 1943 when Pitts was still only eighteen years old.[7] They felt that they were only beginning a line of work that would eventually address the questions of how brain physiology is linked to knowledge.
When Wiener, Bigelow, and Rosenblueth got together with McCulloch and Pitts, in 1943 and 1944, a critical mass of ideas was reached. Pitts joined Wiener at MIT, the worked with von Neumann at the Institute for Advanced Study after the war. By the time this interdisciplinary cross-fertilization was beginning, the ENIAC project had progressed far enough for digital computers to join the grand conjunction of ideas.
A series of meetings occurred in 1944, involving an interdisciplinary blend of topics that seemed to be coming from subject areas as far afield as logic, statistics, communication engineering, and neurophysiology. The participants were an equally eclectic assortment of thinkers. It was at one of these meetings that von Neumann made the acquaintance of Goldstine, whom he was to encounter again not long afterward, at the Aberdeen railroad station. Rosenblueth had to depart for Mexico City in 1944, but by December, Wiener, Bigelow, von Neumann, Howard Aiken of the Harvard-Navy-IBM Mark I calculator project, Goldstine, McCulloch and Pitts formed an association they called "The Teleological Society," for the purpose of discussing "communication engineering, the engineering of control devices, the mathematics of time series in statistics, and the communication and control aspects of the nervous system."[8] In a word--cybernetics.
In 1945 and 1946, at the teleological society meetings, and in personal correspondence, Wiener and von Neumann argued about the advisability of placing too much trust in neurophysiology. Vin Neumann thought that the kinds of tools available to McCulloch and Pitts put brain physiologists in the metaphorical position of trying to decipher computer circuits by bashing computers together and studying the wreckage,
To von Neumann, the bacteriophage--a nonliving microorganism that can reproduce itself--was a much more promising object of study. He felt that much more could be learned about nature's codes by looking at microorganisms than by studying brains. The connection between the mysteries of brain physiology and the secrets of biological reproduction were later to emerge more clearly from theories involving the nature of information, and von Neumann turned out to be right--biologists were to make faster progress in understanding the coding of biological reproduction than neuroscientists were to make in their quest to decode the brain's functions.
The Macy Foundation, which had sponsored the meetings that led to the creation of the Teleological Society, continued to sponsor free-wheeling meetings. Von Neumann and Wiener were the dramatic costars of the meetings, and the differences in their personal style became part of the excited and dramatic debates that characterized the formative years of cybernetics. Biographer Steve Heims, in his book about the two men--John von Neumann and Norbert Wiener--noted the way their contrasting personae emerged at these events:[9]
Wiener was the dominant figure at the conference series, in his role as brilliant originator of ideas and enfant terrible. Without his scientific ideas and his enthusiasm for them, the conference series would never have come into existence, nor would it have had the momentum to continue for seven years without him. A short, stout man with a paunch, usually standing splay-footed, he had coarse features and a small white goatee. He wore thick glasses and his stubby fingers usually held a fat cigar. He was robust, not the stereotype of the frail and sickly child prodigy. Wiener evidently enjoyed the meetings and his central role in them: sometimes he got up from his chair and in his ducklike fashion walked around the circle of tables, holding forth exuberantly, cigar in hand, apparently unstoppable. He could be quite unaware of other people, but he communicated his thoughts effectively and struck up friendships with a number of the participants. Some were intrigued as much as annoyed by Wiener's tendency to go to sleep and even snore during a discussion, but apparently hearing and digesting what was being said. Immediately upon waking he would often make penetrating comments.
In 1944 and 1945, Wiener was already thinking about a scientific model involving communication, information, self-control--an all-embracing way of looking at nature that would include expplanations for computers and brains, biology and electronics, logic and purpose. He later wrote: "It became clear to me almost at the very beginning that these new concepts of communication amd control involved a new interpretation of man, of man's knowledge of the universe, and of society."[10]
Wiener was convinced that biology, even sociology and anthropology, were to be as profoundly affected by cybernetics as electronics theory or computer engineering; in fact anthropologist Gregory Bateston was closely involved with Wiener and later with the first AI researchers. While Shannon published information theory, and von Neumann pushed the development of computer technology, Wiener retreated from the politics of big science in the postwar world to articulate his grand framework.
After the war, as the plans for the Institute for Advanced Study's computer proposed by von Neumann were put into action, with Julian Bigelow as von Neumann's chief engineer on the project, and as Mauchly and Eckert struck out on their own to start the commercial computer industry, Wiener headed for Mexico City to work with Rosenblueth Then, in the spring of 1947, Wiener went to England, where he visited the British computer-building projects, and spoke with Alan Turing.
When he returned to Mexico City, Wiener wrote his book and decided to title it and the new field Cybernetics, from the Greek word meaning "steersman." It was subtitled: or Control and Communication in the Animal and the Machine. Cybernetics was the description of a general science of mechanisms for maintaining order in a disorderly universe, the process for steering a course through the random forces of the physical world, based on information about the past and forecasts about the future.
When a steersman moves a rudder, the craft changes course. When the steersman detects that the previous change of course has oversteered, the rudder is moved again, in the opposite direction. The feedback of the steersman's senses is the controlling element that keeps the craft on course. Wiener intended to embed in the name of the discipline the idea that there is a connection between steering and communication. "The theory of control in engineering, whether human or animal or mechanical," he stated, "is a chapter in the theory of messages."[11]
The mathematics underlying the steering of rudders or antiaircraft guns and the the steering of biological systems was the same--it was a general law, Wiener felt, like the laws of motion or gravity. Wiener's intuitions turned out to be correct. Communication and control, coding and decoding, steering and predicting, were becoming more important to physicists and biologists, who were interested in phenomena very different from guns or computing machines.
In the late 1040s, another new category of interdisciplinary theorists who would come to be known as molecular biologists were beginning to think about the coding mechanism of genetics. Even the quantum physicists were looking into the issues that were so dear to Wiener, Bigelow, and Rosenblueth. It looked as if Wiener might be onto an even more cosmic link between information, energy, and matter. A scientific watershed was imminent, and many if his colleagues were expecting more major breakthroughs from Wiener. By the fall of 1947, prior to its 1948 publication, his book on cybernetics was making the rounds of government and academic experts in manuscript form.
Robert Fano, a professor of electrical engineering who eventually became head of the electrical engineering department at MIT and administrative leader of MIT's pioneering computer project known as MAC, witnessed some strange behavior on Wiener's part around that time, behavior that Fano later had cause to remember when Claude Shannon published his work. Fano was working on his doctoral thesis in electrical engineering. From time to time, Wiener would walk into the student's office, inform him rather cryptically that "information is entropy," and walk out without saying another word.[12]
By the end of 1946, Wiener had reached a decision that had nothing to do with the cold formalisms of mathematics, a decision that distinguished him in yet another way from his weaponry-oriented colleague. Renouncing any future role in weapons-related research, Wiener deliberately removed himself from the hot center of the action in the development of computer technology (as opposed to cybernetic theory) when he stated: "I do not expect to publish any future work of mine which may do damage in the hands of irresponsible militarists."[13] Fortunately for Wiener, and for the scientific world, the implications of his discoveries were not limited to military applications. It quickly became evident that weapons were not the only things of interest that were built from communication and control codes.
By the late forties and early fifties, the atmosphere was crackling with new scientific ideas having to do with what nobody yet called information theory. The quantum physicist Erwin Shroedinger gave a famous lecture at Cambridge University in 1945, later published, on the topic "What is Life?" One of the younger physicists in the audience, Francis Crick, decided to switch to biology, where the most crucial decoding problem in scientific history was waiting for him. Von Neumann turned out to be right in his dispute with Wiener--the bacteriophage, not the nervous system, was the subject of the next great decoding.
Von Neumann's ideas about self-reproducing automata--patterns complex enough and highly ordered enough to direct their own replication--seemed to point toward the same idea. Something about order and disorder, messages and noise, was near the heart of life. The manipulation of information looked like something more like a game mathematicians play, even more than a capability of machines. Information, in a way that was not mathematically demonstrated until Claude Shannon's 1948 publications, began to look like a reflection of the way the universe works. The whole idea was a wrenching of mind-set, at first for scientists, then for many others.
At the beginning of the twentieth century, scientists saw the universe in terms of particles and forces interaacting in complicated but orderly patterns that were, in principle, totally predictable. In important ways, all of the nonscientists who lived in an increasingly mechanized civilization also saw the universe in terms of particles and forces and a clockwork cosmos. Around sixty years ago, quantum theory did away with the clockwork and predictability. Around thirty years ago, a few people began to look at the world and see, as Norbert Wiener put it, "a myriad of To Whom It May Concern messages."
The idea that information is still a fundamental characteristic of the cosmos, like matter and energy, is still young, and further surprise discoveries and applications are sure to pop up before a better model comes along. Before the 1950s, only scientists thought about the idea that