Harry Braverman

Automation: Promise and Menace

(October 1955)

From American Socialist, October 1955.
Copied from the American Socialist Archive in ETOL.
Marked up by Einde O’Callaghan for the Marxists’ Internet Archive.
Proofread by Chris Clayton (July 2006).

WE ARE on the threshold here in America of a remarkable leap in the productive process. Automation, the name given to a generalized theory and practice of semi-automatic production, planning and control, is a genuine innovation, not merely ‘a little more of the same.’ It is already on its way and will be broadened in scope at a rapid pace. It will have smashing consequences upon economy.

Like all technical innovations worthy of the name, automation is a way of producing more, better, with less labor-time. That can spell trouble for our economic system. For capitalism tends to reverse the normal ways of thinking about economic problems. What comes to everybody’s mind when he hears that we are piling up large inventories of goods, if not ‘trouble ahead’? And if we are very low on goods, that is taken to mean that we are in good shape. Because it is an economy of production for profit rather than for use, it thrives on scarcity, and collapses as a result of the very situation that men should be happiest about: what is called a ‘glut’ of the things we consume.

Some have obtusely refused to see anything new in automation. At a recent discussion sponsored by the League for Industrial Democracy, Sol Barkin, research director for the CIO Textile Workers Union, scoffed at the whole thing, and put the finger on ‘rising productivity’ as the real menace; but automation is another means for raising productivity, unmatched in human history. Others have tried to play down automation as something that is not new, the principles of which have been known for hundreds of years. This line is the one most often taken by industry spokesmen trying to allay fears.

IT IS true that the basic principle of automation has been in regular use in industry at least since James Watt invented the flyball governor in 1788. Your home thermostat, the regulating apparatus of a refrigerator, or the steering mechanism which ships have used for years are other long-standing examples. But – and this is the important point – while we may have constructed a number of self-regulating devices in the past centuries, it is a far cry from that to devising an analytic theory which synthesizes the known facts and guides the construction of new devices. And it is a still further leap to combine this idea with electronic machines which can digest, store and act upon information with lightning speed, and then to harness mechanical muscles to these electronic brains – as one journalist has put it.

In the earliest steam engines, James Watt incorporated the following device: The shaft of the engine was made to turn the shaft of a governor, which consisted of two heavy steel balls mounted at the ends of whirling arms, which were in turn connected back to the throttle valve regulating the admission of steam into the engine. As the speed of the engine mounted, the whirling arms flew outward further and further by centrifugal force. But as they flew outward, the lever to which they were connected closed the throttle valve and slowed the machine down; then, as it slowed down, the arms by falling inwards opened the throttle valve again and increased the speed of the engine. By varying the length of the arms, the weight of the flyballs, and the stops which prevented the mechanism from going too far in either direction, the limits of speed could be set.

Now, that is a true example of automation, in spite of the fact that it is simple, because it controls and regulates itself automatically within a closed system; a portion of the energy which it produces is fed back into the system to guide its operation. That is why the term feedback has come to represent a crucial concept in automation. The modern automation systems are no longer so simple as the flyball governor or the thermostat. They depend upon several mechanisms which were developed during the war: the computers or data-processing-and-storing machines, and the servo-mechanism, or devices which transmit signals and amplify them into power to operate the motors and connecting rods which work the machine itself.

WITHOUT going into extended technical detail about these new machines, this much should be made clear: The key fact which gives these control mechanisms their usefulness and superiority is the speed of electrical impulses and the tremendous number of them which can be harnessed in a single instrument by modern electronic science. The governor, by mechanical pushing and pulling of a lever, can transmit a single signal to the engine, and that very slowly. But electronics can send millions of controllable impulses darting with almost the speed of light, thus making possible self-regulating systems of the greatest complexity and rapidity, approximating the very simple operations of the human brain.

What can these machines do? Well, let’s look at what they are doing. In the clerical field, the machine is being put into use at a rapid rate, because once the electronic computer is built, very little need be done to adapt it to office work. An IBM calculator can do over 32,000 multiplication problems of two figures with 13 digits each in one second. In a recent test, a clerical worker needed 540 seconds to do only one such problem, and got the wrong answer! Commonwealth Edison of Chicago will get an IBM of the ‘700 series’ soon which will calculate and print the bills for 1,800,000 customers, use 270 clerks where 470 are now employed, and take only half the time. It will store a ‘memory’ of the past record of every customer, and if the current bill seems too low or too high, it will ask for a new meter reading.

INSURANCE companies, utilities and companies with large payrolls find these machines extremely useful; all large concerns find them an unimaginable labor-saver in planning operations, keeping track of inventories, etc. Department stores will use them to assemble information from cash registers all over the store, or a system of stores if so desired, every day at the close of business, and deliver, with lightning rapidity, a summary of sales in every item handled by the store, and also check these sales against inventories, print and deliver purchase orders and replenish depleted stock where needed, and do everything but roll the shutters down and sweep the floor. The giant data processors are in use in some dozens of the largest corporations already, such as US Steel, General Motors, Metropolitan Life, General Electric, and the manufacturers of the computers, IBM, Remington Rand and others, are preparing and putting on the market smaller machines which will be suitable for all but the smallest firms.

Continuous-flow production processes, especially where liquids or granular solids are involved, are easily regulated by the flow meters and feedback mechanisms of a simple sort which these processes require. That goes for oil, chemicals, flour milling, and so on. An average oil refinery which would employ 800 people would need only 12 people, if instrumentation were used to the fullest. Many refineries are already operating on nearly that basis, and it is doubtful that new refineries will be built otherwise than with complete automation. A Bedford, Ohio, chemicals plant which produces a monthly quota of 650,000 pounds of jellied gasoline, better known as napalm, requires only four men and a supervisor for maintenance on its 10,000 square feet of plant area, and production costs are less than half of the conventional method. The production of electric power by automated atomic-power reactors is in the works: An example is the plant being built in Pittsburgh which will require only six men to furnish electricity for the entire city.

IN THE field of mass production of standardized units, automation is being rapidly introduced into the radio, television, electronics, and other industries. They now have what they call the ‘printed circuit,’ which obviates the need for much hand wiring and soldering. The wiring is etched right on the chassis mount. Admiral Television installed one of the first of the robot production lines; now they are becoming quite common. Philco Radio has a semi-automatic setup that eliminates 37 out of every 40 solderers. A radio assembly line is geared to produce 1,000 radios a day with only two workers; a standard assembly line needs 200 workers for the same output and, by all accounts, the products are more uniform, the connections are trouble-free, the mechanisms resist extremes of temperatures and humidity, they can be engineered with greater freedom than a hand-assembled job, and, of course, the costs are lower, although the price isn’t necessarily.

In auto, the labor savings are prodigious. Ford’s automatic engine plant at Cleveland turns out twice as many engines as an old-style plant with one-tenth the manpower, and the blocks go through in 15 minutes instead of the nine-hour old process. They have a machine in one passenger-car plant, which formerly employed 36 men to feed fenders into a conveyor for spray painting, which automatically feeds six sets of fenders to a fast-moving merry-go-round where different colors of paint are applied at the same time. One worker runs the whole show. Nash machines cylinder heads by a method which reduces man-hours to only 20 percent of the old method, and Pontiac boasts that its 1955 engines are being produced by a plant which ‘surpasses any other plant in the world’ for auto-equipment. All in all, it is estimated that a fully automated automobile industry will require only 200,000 workers to produce the present output of a million auto workers.

Now all of these instances refer to pretty much repetitious types of work, but the automatic machinery of today is going to begin to replace the skilled craftsman, with long years of apprenticeship behind him, too. The Massachusetts Institute of Technology has adapted three servo-mechanisms and a computing system under feedback control to run the three basic motions of a universal milling machine, and that machine can turn out work that it generally takes a skilled machinist to do, simply by reading a punched tape which is fed into a tape reader. The control equipment can run one machine or a battery of machines, and they don’t have to be there in the room either; they can be in another plant, or in another city, for that matter.

AMONG the first to feel the impact of the automation revolution will be the grouping in the economy to which the Marx-refuters have pointed with such a glow of pride: the so-called new middle class. The clerical workers are not really a middle class in the economic sense of the term. They constitute, by virtue of their status as employees – and lower-paid employees at that – a portion of the working class doing a different kind of a job from the overalled worker. In recent years, this grouping has, due to changes in the technical process and to a greater amount of office routine, grown tremendously – from 5 million in 1940 to over 8 million in 1954, or from 11 per 100 industrial workers to 16 per 100. Now that trend will be sharply reversed. Also to be hard hit is the giant layer of semi-skilled workers – the biggest portion of the country’s working class – the class of workers created by the Industrial Revolution and the rise of mass-production methods. A great mathematician and scientist who, probably more than any other single person deserves the name of the father of this new branch of science, Norbert Wiener, put it this way in his book Cybernetics:

‘Perhaps I may clarify the historical background of the present situation if I say that the first industrial revolution, the revolution of the ‘dark satanic mills,’ was the devaluation of the human arm by the competition of machinery. There is no rate of pay at which a United States pick-and-shovel laborer can live which is low enough to compete with the work of a steam shovel as an excavator. The modern industrial revolution is similarly bound to devalue the human brain at least in its simpler and more routine decisions. Of course, just as the skilled carpenter, the skilled mechanic, the skilled dressmaker have in some degree survived the first industrial revolution, so the skilled scientist and the skilled administrator may survive the second. However, taking the second revolution as accomplished, the average human being of mediocre attainments or less has nothing to sell that is worth anyone’s money to buy.’

That’s pretty strong, but it was written by a man in a position to know what he’s talking about in this field, and was written, moreover, almost eight years ago, before the industrial application of his science was rushing upon us. And it is rushing pretty fast today, too. A booklet which the General Electric Corporation distributed to its employees called Automation – Friend or Foe? told them:

‘The employer must automate to stay alive ... It is imperative ... that he remove from his payroll any substantial surplus of employees not needed.’

When once automation is introduced by one large manufacturer in an industry, it must rapidly be introduced by his competitors on pain of sudden death. You don’t have to be an expert in monopoly, oligopoly, the theory of imperfect competition or anything else to figure that out; it’s an elementary fact of our business system. And, since automation is very often surprisingly cheap, and can also often plied to existing equipment, the pace is being rapidly.

In 1940, factory sales of data-process equipment were zero, in 1953, $25 million, and in 1960 expected to be $500 million. Instruments for industrial control sold to the tune of only $3 million in 1940, $65 million in and estimates for 1960 are $150 million.

These figures would not mean much without a yardstick, so let me tell you what Professor Wassily of Harvard University figured out for the Scientific American of September 1952: If all the new plants 1950 had been automated, that would have cost an additional $600 million in equipment; actually, about $67 million was spent in that year, and this was enough to automate about 10 percent of all plants being built. Since the spending has risen a great deal since then, we may be sure that a substantial percentage of all new plants being built are automated. It is estimated that, if the trend continues, American industry will be fully automated to the reasonable limit within ten years, meaning, according to the that one man will do at least the work that five do now.

WE COME now to the big question: What will effect of such automation upon our economy. The air is thick with reassurances. In substance, the argument of the more conscientious apologists is something like this: ‘We are willing to admit that there will be placements and hardships, but in the long run, more jobs will be created. We’ve had this before, and we’ve seen that, in the end, technological improvement led to more jobs instead of fewer.’

There are a number of different angles to this problem, and I propose to examine several of them. The placement of workers due to new technologies is made up, in part, by increased employment in the goods industries where those new machines are being built. I emphasize the ‘in part’ for this reason: If it took as much labor to build the machinery as that machinery displaced, then the machine would not be economical and would not be used. There is no sense in paying more in wages to build a machine than you can save by installing it; you lose on the deal.

During the first Industrial Revolution and all later phases up to the recent period, the capital-goods requirements were massive; entire new industries brought into being. But the remarkable thing about automation is that it saves labor on a greater scale than ever before at the cost of comparatively little money, and it will therefore not boost employment in the goods department of capitalism very much. Wassily Leontief of Harvard, whose work I mentioned before, calculated the approximate costs of automating capital equipment. He figured the costs as a percentage of cost of new equipment; thus if a chemical plant can be built for $900,000 and the same equipment can be automated for another $100,000, then the cost of automation is 10 percent of the total cost of the new installation. His figures show that automation for a pulp mill three percent of the total cost of the mill; in chemicals and oil refining, between 5 and 6 percent; petroleum refining, 3-5 percent; packaged foods, 3-5 percent; mining processing of ores, 2-5 percent; rubber, 5 percent; and carpets, 2-3 percent. The highest figure on his list was for meat-packing, which runs 19 percent, and the lowest for rayon and rayon yarn, 1-2 percent. The average figure was the amazingly low one of only 6 percent. Leontief comments on this: ‘The mechanization of the nineteenth century required heavy capital investment and proceeded slowly; the new technology, unhampered by vast capital requirements, can be introduced at a faster pace.’

Hopes for the growth of vast new capital-goods industry like those of a century ago, and of a stimulus to employment in those industries and in the secondary fields boosted by them, are bound to be disappointed. A basic difference with the first industrial revolution appears as soon as we begin to look into the problem. Not only that, but the electronic industries which produce the new equipment are themselves singularly susceptible to automation. A Department of Labor study informs us:

‘Electronics output in 1952 was 275 percent higher than in 1947 but was produced by only 40 percent more workers. Output per man may rise even faster during the next few years as a result of improvements in manufacturing techniques ... These trends toward ‘automation’ may result in the greatest reduction in man-hours in the industry’s history during the next years.’

The IUE-CIO has estimated that 110,000 workers in electrical industry lost their jobs in the 21 months ending last November 30 primarily because of the increase in automatic equipment. These facts are of the greatest importance, for they mean that we must not look to any hiring spree in the main new industry which is being produced by the new industrial revolution.

NOW WE come to the second aspect of this matter. Technological innovations can be of two major kinds: labor-saving capital-goods equipment, or new consumer products. The lathe or milling machine are examples of the first; the automobile and radio examples of the second. The economists have made much of the fact that, while Marx examined, in detail, the effects of machines that increase productivity upon the economy, he paid scant on to inventions that place new consumer products the market. And they have boasted that the technological revolution of the early 1900’s in America was concerned largely with the development of new products, which stimulated consumer demand and, by a process like that of spreading ripples throughout the economy, gave a boost to every aspect of production and consumption. Their boast had some substance to it, and this fact helped to give the American economy new vigor for a while. But, as you can clearly see from the broad facts without my citing statistics or quotations, that is not involved in the present industrial revolution. This revolution is simply bearing down on reducing the hours of labor that go into the existing consumption goods. Therefore, the big boosting factor of the early part of the century is far less, not more, operative.

Another feature of the past industrial revolution during the heyday of the first Henry Ford was the multiplication of the service industries. Here we must get a little background: When David Ricardo, the great British classical economist, first turned his attention to the effect of labor-displacing machinery, he noted at once that if an English manufacturer took the workers who were thrown out of jobs by a new textile machine, for example, and employed them as servants on his estate – gamekeepers, butlers, coachmen, etc . – using the money he saved by firing them to rehire them for the purpose of making his personal life more luxurious, then the problem of technological unemployment would be solved. This axiom noted by Ricardo came into play in the early twentieth century in the United States in a peculiar way. The servant class didn’t increase, it declined, but the number of those engaged in servant activities for society as a whole grew tremendously. I am speaking of the vast expansion of the service industries: the staffs of luxury hotels and resorts, the vast army of attendants in the gas stations and auto repair shops of the nation, the personnel of the radio stations, the new multiplicity of agents and dealers of all kinds, etc. As great new avenues of consumption were opened up for the capitalist and middle classes, and as these avenues siphoned away a portion of what would otherwise have become unemployed capital faced on the other side by unemployed labor, it had a huge economic effect.

In the seven fat years from 1923 to 1929, at a time when industrial employment was slowly falling, employment in the service industries climbed by the phenomenal figure of 70 percent, and for a time, kept the pressure off the labor market and off wages in the industrial field. The automobile had the largest role in this development. But in the new industrial revolution, that kind of growth is not to be expected, because it is not inherent in the technical nature of the present revolution.

IT IS clear that capitalism can assimilate rapid technological advances only if it can expand greatly, to invest new hoards of capital and thus re-employ some of the displaced workers. During the eighteenth and nineteenth centuries capitalism was growing very rapidly, not only by increasing the market in its own arenas, but also at the expense of old systems of production, taking over new areas of its own nations and of other nations. For example, in the period from 1800 to the present, when the farming class, which used to be 80 percent of the population and was semi-independent of capital, was reduced to only 20 percent of the population, capitalism was in the course of this process acquiring great new areas for investment at the expense of an older economic form. One of the special reasons for the exceptional vigor of American capitalism was that it was able to repeat this process several times over. It destroyed and then re-created through immigration and a land abundance, a series of such farming classes, and so was able to have its cake and eat it too, a number of times. But that day is over now.

Capitalism expanded also at the expense of the colonial lands. The economist Fritz Steinberg has calculated that where, when Marx and Engels wrote the Communist Manifesto, only 10 percent of the world’s population lived directly under the capitalist system, by 1914 this had increased to 30 percent. Imagine if capitalism had been compelled to keep increasing its investments and re-employing slack labor within that same 10 percent of the world! It would have been impossible. But, since 1914, not only hasn’t the capitalist 30 percent grown, it has even shrunk a bit, and, on top of that, 35 percent of the earth’s population has been permanently taken out of the orbit of capitalist possibilities. The contrast is devastating and shows that world capitalism will develop acute disorders in trying to assimilate the effects of this new industrial revolution.

I have been trying to reply to the breezy claim that the capitalist system will take the new revolution in its stride as it took the old machines – by expanding and readjusting. But, in order to make the comparison, I have accepted this claim – that the last industrial revolution was surmounted by capitalism – as good coin. I fact, the tale is a different one. The adjustments were very badly made, and involved huge industrial reserve armies. And finally, the time came when the system could not adjust at all.

We had a big upsurge of productivity early in the century, the hucksters tell us, and we had more employed all the time, and better jobs, too. But then what happened? The great expansion of production, with which consumption wasn’t permitted to keep up by workings of the capitalist system, threw us into the most devastating depression in history, with 10 million unemployed even at the end of it – when a war economically pulled us out. That’s where Henry Ford’s ‘job-producing’ production line put us in the end; an the part of the story that we are never told when we are referred to the wonder-working miracles of technology in our past economy.

ALL THE comparisons with an earlier day point unerringly to a single conclusion: If capitalism in the United States succeeds in accomplishing a more complete automation – which is the logical end which industry is heading – the effects will be devastating. American economy will have worked itself into an impasse. And the great promise of automation in a society, which must be left to another time for discussion must surely be clear to all who understand that an economy which produces for use instead of profit can suffer from our present embarrassment of riches.


Last updated on 19.7.2006