I came to Norbert Wiener by the back door. I was reading James Gleick's The Information, a history of the idea that messages can be measured, and I noticed, as one notices which child a parent prefers, that the book had a favourite. It was Claude Shannon. Shannon was the Bell Labs engineer who showed, in 1948, how much information can be sent through a noisy wire. Wiener was the MIT mathematician who published a larger and stranger book in the same year, and who appears in Gleick's pages, as I read them, with respect and without affection. I do not hold this against Gleick. Shannon's work is the better-behaved of the two, and the modern world runs on it. But when I am told that one runner is the obvious winner, I want to know about the other. This essay is the result.
It is a manifesto about a manifesto. Wiener's Cybernetics is a manifesto, and I mean the word as praise. A manifesto names a field, draws its boundary and announces what the field will do. A book that does these three things is not merely reporting on the world. It is proposing a new way to divide it. I think the proposal was largely right, and I think its origin lay in an activity that I find hard to look at. I intend to say both.
A word about method first. I have a habit I call "and, also." When two statements are true and seem to compete, I write the first, then "and" the second, and I do not put "but" between them. The word "but" is how we usually excuse ourselves. "He was a brilliant mathematician, but his work was used in war" invites the reader to subtract one fact from the other and keep the remainder. I would rather add. The sum is less comfortable and more accurate.
Cybernetics: Or Control and Communication in the Animal and the Machine appeared in 1948. Wiener took the word from the Greek kybernetes, which means steersman. A steersman does not push a ship toward a harbour. He observes where the ship is going, compares that with where it ought to go, and moves the rudder to reduce the difference. The comparison and the correction are the whole art. Wiener's claim was that this arrangement, which he called feedback, is found in a thermostat, in a gun that follows an aircraft, in a hand reaching for a cup, and in the nervous system that guides the hand. Feedback means that a system's output is measured, that the measurement is returned to the system as an input, and that the difference between what happened and a target governs what happens next. In a thermostat the target is a number that someone set on a dial. Nothing in the definition requires that the system understand what it is doing, or want anything. It requires a measurement, a target and a correction.
The idea is easier to feel than to define. Wiener's own example, as I remember it, was a person picking up a pencil. The person does not command the hand to travel a fixed distance. The eye reports how far the hand still is from the pencil, and the arm shortens that gap until it is zero. He noted that a patient with a certain injury to the cerebellum, the part of the brain that helps to coordinate movement, overshoots, and then overshoots again in the other direction, because the correction arrives late and too large. The same failure appears in an ordinary shower. You turn the tap, nothing happens for several seconds, you turn it further, and then the water is too hot. The fault is in the delay between the correction and its report. Once you have seen this, you see it in a person adjusting a habit, and in a gardener watering a plant that shows its answer a week later. It does not change what you can see. It changes what you look for.