relocation of industry, shifts in market demands, and other changes in the structure of the economy. Full employment, in turn, is the result of a proper mix of monetary, fiscal, tax, labor, and other related policies, legislative as well as administrative.

2. There is no clarity with respect to the fact that the effect of technological change (including automation) on different industries is not uniform. Thus, for example, the office, computing, and accounting machines industry can anticipate a major increase in employment during this decade; the primary metals, textile products, and printing and publishing industries, a major decrease during the same period. Similarly, while modest increases will be registered by the electrical machinery, communications, miscellaneous business services, and amusement and recreation industries, moderate decreases will at the same time be registered by the nonelectrical machinery, transportation equipment, railroad transportation, banking, and insurance carrier industries. Other industries will show either slight or no employment change during this decade.

3. There is no clarity with respect to the fact that a high increase in productivity in a given industry does not necessarily lead to a downward trend in employment. In fact, the degree of automation or mechanization in an industry is not the sole and may not be the major determinant of employment trends. This is so because an increase in productivity usually represents the combined effect of various elements-worker's skill, managerial skill, technological change (including automation), capital investment per worker, capacity utilization, layout and materials flow, as well as labor-management relations. Thus, in the period of 1947-61, certain major industries significantly increased both aggregate employment and productivity per person:

4. There is no clarity with respect to the fact that while training and retraining is not a new dimension in our national policy, it is only part of a national labor market policy. At its minimum, such a policy should include current labor market information services, an "early warning" system of employment changes, career guidance and counseling services, an educational system (vocational as well as academic) directed to proper manpower resource utilization, a national system of placement services as well as a program of training and retraining for unemployed and underemployed.

5. There is no clarity with respect to the fact that the unemployment due to technological change (including automation) may result in either unemployed with unused skills or unemployed with obsolescent skills. While, generally speaking, a proper national labor market policy would properly care for both significantly, different emphasis is necessary with respect to alleviating the hardships resulting from each kind of unemployment.

In conclusion, what is lacking, generally speaking, is a clear, unified understanding of the employment implications of technological change (including automation) and the specific appropriate methods of adjustment to such changes. No one in Government-nothwithstanding the noteworthy initial efforts in this direction of the U.S. Department of Labor's Office of Manpower, Automation, and Training—is looking sufficiently ahead. There is only sparse realization that we are on the threshold of momentous technological changes. No one has assessed in depth its implications on the job and off the job. Therefore, no one in Government can properly plan ahead with respect to these matters.

Thus, while the fullest support should be given to the proposed amendments to the Manpower Development and Training Act (embodied in H.R. 8419), additional legislation is necessary. Such legislation should be based upon thorough examination into the full implications of technological change (including automation) upon the labor force and its structure, and the necessary methods to adjust to such change. The necessary body of data to support such additional legislation does not at present exist.

With this in mind, one should welcome President Kennedy's request on July 22, 1963-now embodied in Senate Joint Resolution 15-to establish a Presidential Commission on Automation. The proposed Commission would recommend, on the basis of comprehensive review of all relevant factors, the specific legislative and administrative steps which should be taken by the Federal Government in order to meet its responsibility of assuring an environment conducive to technological change while alleviating the dislocation resulting from the application of technological change. In addition, this proposed Commission would identify and describe the major types of worker displacement, both technological and economic, likely to occur during the coming 10 years and the social and economic effects of these developments on our economy, our manpower, our communities, and our social structure. The establishment of this proposed Commission, as an important instrument of national policy, is long overdue.

May I thank this subcommittee for its most respectful consideration.

[From "Cybernation-the Silent Conquest" (Center for the Study of Democratic Institutions, 1962)]

CYBERNATION-THE SILENT CONQUEST

(By Donald N. Michael)

Both optimists and pessimists often claim that automation is simply the latest stage in the evolution of technological means for removing the burdens of work. The assertion is misleading. There is a very good possibility that automation is so different in degree as to be a profound difference in kind, and that it will pose unique problems for society, challenging our basic values and the ways in which we express and enforce them."

1This paper makes the following assumptions in looking on the next 20 years or so: (1) International relations will derive from the same general conditions that pertain today: (2) the weapons systems industries will continue to support a major share of our economy: (3) major discoveries will be made and applied in other technologies, including psychology and medicine; (4) trends in megalopolis living and in population growth will continue and (5) no major shifts in underlying social attitudes and in public and private goals will take place.

In order to understand what both the differences and the problems are, and even more, will be, we have to know something of the nature and use of automation and computers. There are two important classes of devices. One class, usually referred to when one speaks of "automation," is made up of devices that automatically perform sensing and motor tasks, replacing or improving on human capacities for performing these function. The second class, usually referred to when one speaks of "computers," is composed of devices that perform, very rapidly, routine or complex, logical, and decisionmaking tasks, replacing or improving on human capacities for performing these

functions.

Using these machines does not merely involve replacing men by having machines do tasks that men did before. It is, as John Diebold says, a way of "thinking as much as it is a way of doing. *** It is no longer necessary to think in terms of individual machines, or even in terms of groups of machines; instead, for the first time, it is practical to look at an entire production or information-handling process as an integral system, and not as a series of individual steps." For example, if the building trades were to be automated, it would not mean inventing machines to do the various tasks now done by men; rather, buildings would be redesigned so that they could be built by machines. One might invent an automatic bricklayer, but it is more likely that housing would be designed so that bricks would not be laid. Automation of the electronics industry was not brought about through the invention of automatic means for wiring circuits but through the invention of essentially wireless, i.e., printed, circuits (though today there are automatic circuit wirers as well).

The two classes of devices overlap. At one pole are the automatic producers of material objects and at the other, the sophisticated analyzers and interpreters of complex data. In the middle zone are the mixed systems, in which computers control complicated processes, such as the operations of an oil refinery, on the basis of interpretations that they make of data automatically fed to them about the environment. Also in this middle zone are those routine, automatic, data-processing activities which provide men with the bases for controlling, or at least understanding, what is happening to a particular environment. Processing of social security data and making straightforward tabulations of census information are examples of these activities.2

Cybernated systems perform with a precision and a rapidity unmatched in humans. They also perform in ways that would be impractical or impossible for humans to duplicate. They can be built to detect and correct errors in their own performance and to indicate to men which of their components are producing the error. They can make judgments on the basis of instructions programed into them. They can remember and search their memories for approppriate data, which either has been programed into them along with their instruc

In order to eliminate the awkwardness of repeating the words "automation" and "compaters" each time we wish to refer to both at the same time, and in order to avoid the semantic difficulties involved in using one term or the other to mean both ends of the Continuum, we invent the term "cybernation" to refer to both automation and computers. The word is legitimate at least to the extent that it derives from "cybernetics." a term invented by Norbert Wiener to mean the processes of communication and control in man and machines. He derived it from the Greek word for "steersman." The theory and practice of cybernetics underlie all systematic design and application of automation and computers.

tions or has been acquired in the process of manipulating new data. Thus, they can learn on the basis of past experience with their environment. They can receive information in more codes and sensory modes than men can. They are beginning to perceive and to recognize.

As a result of these characteristics, automation is being used to make and roll steel, mine coal, manufacture engine blocks, weave cloth, sort and grade everything from oranges to bank checks. More versatile automatic fabricators are becoming available, too:

U.S. industries announced *** that it had developed what was termed the first general-purpose automation machine available to manufactuers as standard "off the shelf" hardware. * * * The new machine, called a TransfeRobot, sells for $2,500. * ** The Westclox Co., of La Salle, Ill., has been using a Transfe Robot to oil clock assemblies as they pass on a conveyor belt. The machine oils eight precision bearings simultaneously in a second. At the Underwood Corp. typewriter plant in Hartford, the robot picks up, transfers, and places a small typewriter component into a close-fitting nest for an automatic machine operation. In an automobile plant, the device feeds partly fabricated parts of a steering assembly to a trimming press and controls the press. The device consists basically of an arm and actuator that can be fitted with many types of fingers and jaws. All are controlled by a self-contained electronic brain.

At the other end of the continuum, computers are being used rather regularly to analyze market portfolios for brokers; compute the best combination of crops and livestock for given farm conditions; design and "fly" under typical and extreme conditions rockets and airplanes before they are built; design, in terms of costs and traffic-flow characteristics, the appropriate angles and grades for complex traffic interchanges; keep up-to-date inventory records and print new stock orders as automatically computed rates of sales and inventory status indicate. Computers have also been programed to write mediocre TV dramas (by manipulating segments of the plot), write music, translate tolerably if not perfectly from one language to another, and stimulate some logical brain processes (so that the machine goes about solving puzzles and making mistakes in the process in the ways people do). Also, computers are programed to play elaborate "games" by themselves or in collaboration with human beings. Among other reasons, these games are played to understand and plan more efficiently for the conduct of wars and the procedures for industrial and business aggrandizement. Through such games, involving a vast number of variables, and contingencies within which these variables act and interact, the best or most likely solutions to complex problems are obtained.

The utility and the applicability of computers are being continually enhanced. For example, after a few hours of training, nonspecialists can operate the smaller computers without the aid of programers simply by plugging in prerecorded instruction tapes that tell the computer how to do specific tasks. Instruction-tape libraries can supply preprogramed computer directions for everything from finding cube root of a number to designing a bridge. When the machine is through with one task, its circuits can be easily cleared so that a new set of preprogramed instructions can be plugged in by its businessman operator.

the

But the capabilities of computers already extend well beyond even these applications. Much successful work has been done on computers that can program themselves. For example, they are beginning to operate the way man appears to when he is exploring ways of solving a novel problem. That is, they apply and then modify, as appropri

ate, previous experiences with and methods of solution for what appear to be related problems. Some of the machines show originality and unpredictability. To take one example from a recent paper of Norbert Wiener:

The present level of these learning machines is that they play a fair amateur game at chess but that in checkers they can show a marked superiority to the player who has programed them after from 10 to 20 playing hours of working and indoctrination. They thus most definitely escape from the completely effective control of the man who has made them. Rigid as the repertory of factors may be which they are in a position to take into consideration, they do unquestionably-and so say those who have played with them-show originality, not merely in their tactics, which may be quite unforeseen, but even in the detailed weighting of their strategy.

Another example of a machine the behavior of which is not completely controllable or predictable is the Perceptron, designed by Dr. Frank Rosenblatt. This machine can learn to recognize what it has seen before and to teach itself generalizations about what it recognizes. It can also learn to discriminate, and thereby to identify shapes similar to those it has seen before. Future versions will hear as well as see. It is not possible to predict the degree and quality of recognition that the machine will display as it is learning. It is designed to learn and discriminate in the same way that it is believed man may learn and discriminate; it has its own pace and style of learning, of refining its discriminations, and of making mistakes in the process.

It is no fantasy, then, to be concerned with the implications of the thinking machines. There is every reason to believe that within the next two decades machines will be available outside the laboratory that will do a credible job of original thinking, certainly as good thinking as that expected of most middle-level people who are supposed to "use their minds." There is no basis for knowing where this process will stop, nor, as Wiener has pointed out, is there any comfort in the assertion that, since man built the machine, he will always be smarter or more capable than it is.

It may be seen that the result of a programing technique of (cybernation) is to remove from the mind of the designer and operator an effective understanding of many of the stages by which the machine comes to its conclusions and of what the real tactical intentions of many of its operations may be. This is highly relevant to the problem of our being able to foresee undesired consequences outside the frame of the strategy of the game while the machine is still in action and while intervention on our part may prevent the occurrence of these consequences. Here it is necessary to realize that human action is a feedback action. To avoid a disastrous consequence, it is not enough that some action on our part should be sufficient to change the course of the machine, because it is quite possible that we lack information on which to base consideration of such an action.

The capabilities and potentialities of these devices are unlimited. They contain extraordinary implications for the emancipation and enslavement of mankind.

The opportunities for man's enhancement through the benefits of cybernation are generally more evident and more expected, especially in view of our proclivity to equate technological advances with progress and happiness. In the words of the National Association of Manufacturers:

For the expanding, dynamic economy of America, the sky is indeed the limit. Now more than ever we must have confidence in America's capacity to grow. Guided by electronics, powered by atomic energy, geared to the smooth, effortless

36-510-64 vol. 1-28