More from 1955... Stop coining words!
Before robots and computers, what nomenclature did we use? This March 1955 Control Engineering article argues against today’s word: automation.
Happy 60th, Control Engineering! Our magazine first published in September 1954. This monthly column in 2014 will resurrect some of our favorite material from the 1954 and 1955 issues. Technologies have progressed, but they continue to pave the way for today’s innovations. Here is a full-length article penned in 1955 by J. J. Brown of Aluminum Fiduciaries Ltd.
We stand today at the beginning of a great new development in technology. This is automatism—control of machinery to bring about fully automatic production of goods and services. For military and economic reasons we should advance in this field as rapidly as possible. Rapid advance in any science is helped by good definitions, clearly understood. These definitions should be existing dictionary definitions, in order that today’s writing on the subject may tie in closely with the writings of the past.
Whenever a new subject appears there is a tendency to coin words. This urge should be suppressed. The automatic production of goods and services is a complex subject, and ill-considered coinages like “automation” only add to the confusion.
The automatic factory is a device capable of producing goods without having any men in series with its machines. It is helpful to think of progress toward the automatic factory as a series of steps. At each step the function of the human being is reduced, first by the arrangement of machines in such a way as to reduce the amount of physical work required, and then later by changers reducing his mental work.
The word for the first step toward automatic production is “mechanization.” This word is defined in the New English Dictionary as “the process of making mechanical.” The modern transfer machine that performs a series of operations on a rough casting is a highly mechanized device. So was Arkwright’s power loom. Other examples are the wire-making and coining machines common as far back as the 14th century.
At a round table on the subject of automation sponsored by Fortune Magazine a year ago last June, I tried to make the point that the dictionary already contains a series of words adequate to describe all known forms of automatic control of processes and machines. But I had little success in persuading the industry members of the round table that we already had perfectly good words. Since most of the men writing about the new mechanized plants are from the industry, the word “automation” is catching on. This is bad. It is already causing unnecessary confusion which will get worse with the passage of time. Occasionally there is a need for new scientific words, but this isn’t one of them.
Automatism is the self-control of machines or operations by any means. The key word here is “control.” It includes also the supervision of processes as seen in the petrochemical and oil industries. In the dictionary “automatism” is defined as the state or quality of being automatic, and “automatic” means self-acting or self-regulating. A machine is an automaton when it supplies entirely by itself the information needed to keep it running satisfactorily. The basic criterion for judging whether or not a machine is an automaton is to ask what would happen if the man nearest it suddenly has a heart attack and dies. If the machine will go on performing its function exactly the same way and indefinitely, it is an automaton. If, on the other hand, its functioning is in any way connected with the actions of a human operator, it is not. Think of the headlight dimmer on your Cadillac—an automaton—and contrast it with the automobile itself— a machine.
Automatism in its most general meaning is what we have when there is a coordination of self-operating machines. Thus in a truly automatic factory there are no men at all in series with the production or transport machines. If in a factory there are any men doing meaningful operations on the production floor, it is not an automatic factory, but a highly mechanized one.
It is important to insist on the essential difference between mechanization and automatism because the economic, social, and design consequences of the two are quite different.
It is true that mechanization, if carried to the point where the last workman is removed from the machines, will turn into automatism. But this is not going to happen. The essential difference between machines and automata is in the question of communication. “How easy is it to tell the device what to do?” If it turns out to be easy, you are dealing with automatism.
The two key words in this field then are “mechanization,” which eliminates some men from the factory floor, and “automatism,” which tends to eliminate them all. It would be elegant and tidy if we could go on to make some such generalization as this: “Mechanization is the process of having machines do physical work formerly done by men; automatism is the process of having machines do mental and control operations formerly done by men; automatism is the process of having machines very difficult to sustain.” Some highly mechanized devices such as adding machines do away with mental rather than physical work. Some highly automatic devices replace manual labor.
There is, however, a degree of truth in the statement that mechanization has more to do with physical work and automatization with mental work. Mechanization is by far the older of the two processes, and has gone through many more stages of development. Today automatism is taking great forward strides. Devices that show higher and higher capabilities and self-control are beginning to appear. For this reason we are beginning to need a new word to cover those automata in which the control function predominates. The word for this is “cybernetics.” For about 25 years French engineers have used the term “l’industrie cybernetique” to describe the branch of automatism that is concerned exclusively with control mechanism. Norbert Wiener of MIT has popularized the term on this continent in three books published since 1948, and it seems to fill a legitimate need.
The objection can be raised that automatism, having been in use long before we knew any methods by which high levels of automatic action be achieved, carries with it the burden of older meanings. One way to answer this objection is to use “cybernetics” to describe areas of automatism where the emphasis is on the control function rather than on the action itself. The vane of a windmill is a good example, and so is an airplane’s automatic pilot. However, automatism has never been a highly popular word; hence its accretions of meaning are slight. The New English Dictionary gives only three examples of its use, all from the 17th and 18th centuries, and all using the word in its philosophical sense as the antonym of “voluntary.”
To sum up, the terminology of automatic production that seems adequate is as follows:
Where there is no program, but repeated action. The control is mostly fixed, as for example in the linked transfer machines:
- Generic noun: mechanization
- Adjective: mechanized
- Verb: to mechanize
Where there is a program of some kind which causes repeated action. The actions repeated are not the same each time because the sensing devices have brought in information to modify the action:
- Generic noun: automatism
- Noun: automaton
- Adjectives: automatic or automatized
- Verb: to automatize
Where records or sensors cause repeated action, but the emphasis is on the control rather than on the output. Here the actions are even less likely to be exactly the same each time, because many sensors have brought in control information to modify them:
- Generic noun: cybernetics
- Adjective: cybernetic
- Verb: to cybernetize
- 2014 Edits by Jordan Schultz, content manager, CFE Media, jschultz(at)cfemedia.com.
- See related stories below.