The machine-tool industry occupies a place of rare importance in the literature of economics. In part that is so because of the things the industry makes. From the machine shops of a nation come the dies that are used to stamp or form nearly every mass-production item, from automobile fenders to soft-drink bottles—as well as the precision-machined goods of the old and new industrial eras, from tank turrets and turbine blades to disk drives for computers. But there is another reason that labor economists, especially those of the left, have concentrated on machine tools: the industry embodies the labor theory of value more fully than any other.
Unionized workers in the auto- and steel-making industries are still the princes of labor, if earnings are the only measure; but theirs is the strength of large numbers, not of great skills. Their ultimate threat is to impede the flow of bodies through the plant gate. But master machinists, through the skills they have developed in years of experience with lathes and cutting tables, possess power independent of their numbers. Employers cannot credibly threaten to bring in a fresh work force, for new hands could not do the job. As other trades have become more and more routine and automated, machinists remained the industrial age’s closest approximation to independent artisans. Although they have never been numerous—there were roughly half a million machinists, broadly defined, in 1980—their fate is disproportionately important to those who study the connection between technological progress and the welfare of the working class.
Before David Noble’s Forces of Production only one other book that I know about underscored the drama and significance of the machine-tool industry. In Player Piano, published in 1952, Kurt Vonnegut described the coming of automation to a factory that closely resembled the General Electric plant in Schenectady where Vonnegut had briefly worked as a publicist. The bright young managers and engineers of Vonnegut’s story, college trained and with clean hands, set about capturing the machinist’s skills on magnetic tape, so as to bring a new age of efficiency to the plant.
They searched for the most accomplished machinist, selected one Rudy Hertz, and attached recording instruments to his lathe. As he worked, the motions he had refined over the years, and from which he made his living, were transferred to the loop of tape. Hence-forth the tape could drive other lathes, as a roll of recorded music could drive a player piano. Long after the transformation was complete and the likes of Rudy Hertz had been reduced to economic and spiritual redundancy, one of the engineers, Paul Proteus, stares at the loop of tape and remembers the time when it was made. Noble includes the passage as an appendix to his book:
Rudy hadn’t understood quite what the recording instruments were all about, but what he had understood, he’d liked: that he, out of thousands of machinists, had been chosen to have his motions immortalized on tape.
And here, now, this little loop in the box before Paul, here was Rudy as Rudy had been to his machine that afternoon—Rudy, the turner-on of power, the setter of speeds, the controller of the cutting tool. This was the essence of Rudy as far as his machine was concerned, as far as the economy was concerned…. The tape was the essence distilled from the small, polite man with the big hands and black fingernails…. Now, by switching in lathes on a master panel and feeding them signals from the tape, Paul could make the essence of Rudy Hertz produce one, ten, a hundred, or a thousand of the shafts.
David Noble’s Forces of Production is an anatomy—more precisely, a pathology—of the real economic transformation that paralleled Vonnegut’s account. Noble, an associate professor of the history of technology at MIT, writes from a Marxist perspective, and the conspiratorial motives of winning the class war that he imputes to his capitalists are the least substantiated parts of his argument, even though they are essential to the point he ultimately wants the book to make. The book’s jacket informs us that Noble has, in addition to a Ph.D., “a certificate in machine-tool fundamentals from the Lowell Institute School,” and he seems to take perverse pride in ostentatiously refusing to explain what the machines he is writing about look like or how they work. By the end of his book, the reader may have developed mental pictures of a ball-end plunge cutter and related items of hardware, but not through any help from Noble.1
But these are mere irritants; despite them, Forces of Production is a prodigious accomplishment. Through the depth of his research and the care of his narration, Noble has produced a detailed, gripping, and convincing work of social history, which adds detail, sinew, and emotion to our understanding of issues that are usually considered only in the abstract.
Noble’s starting point is the machine-tool industry of the late 1940s. Two of its traits were especially significant. One was the control that the machinists, through their indispensable mastery of production skills, exercised over daily operations. Managers might whine, threaten, and cajole in their attempts to set daily quotas, but the machinists had the final word. The other trait was the bitter industrial strife of the postwar years, in which every item in the labor contract, from wage rates to shop-floor control, seemed to be up for renegotiation.
From the managers’ point of view, Noble says, the second characteristic made the first—the machinists’ power—intolerable. Therefore they resolved to remove that power by destroying the skills on which it was based. “The same skills that make production possible…give the workers de facto command of the shop,” Noble writes. “Whether they used that command to increase output…or restrict it was never the main concern of management. The central problem was simply that the power belonged to the work force rather than management.”
Then came the crucial new development: the rise of the postwar defense establishment. Because of military contracting, Noble says, the machine-tool industry was transformed in a way and at a rate that would otherwise have been barely imaginable.
In part, the military exercised its influence merely as a byproduct of its quest for more sophisticated weaponry. To power its most modern aircraft engines or guide its newest missiles, it needed (or at least demanded) parts that were ground to precise tolerances—so much more precise than previous standards that mere men, working at their machines, were thought unable to attain them.2 The military brought something else to the industry: money, in quantities far beyond what the commercial market was willing to risk. The higher standards required engineers to experiment with exotic ways of automating the machine shop, and military contracts financed them as they tried.
Other writers, from Seymour Melman to Lester Thurow, have discussed the impact of military contracting on the private economy, but I know of no one who has explained the process in such precise, clinical detail as Noble does here. The themes that run through his account are illustrated in one of the earliest and most important episodes he describes, the struggle between a businessman named John T. Parsons and the Servomechanisms Lab at MIT.
Even before the newly created Air Force began asking for more precisely machined aircraft parts, there had been attempts to automate certain machining processes. In general, these involved analog methods—a machinist might make a wooden model or template of the item he was to produce, and then use that template as a guide for his cutting tool while shaping the finished metal product. Noble says that these innovations were typically introduced by machinists themselves, who remained in control of the pace and quality of production.
But there had been other early approaches to automation, which reflected a desire for more precise results that digital, rather than analog, control might bring. A roll of player-piano music, in which each punched hole corresponds to exactly one note, is an example of digital control; so was the Jacquard automatic loom, introduced in the early nineteenth century to replace the artisans who had drawn different threads through a loom to produce finely patterned fabrics.
John Parsons, owner of a small business in Michigan that was the nation’s leading producer of helicopter rotor blades, decided during the 1940s to experiment with numerical control of machine tools. (“Numerical control,” abbreviated N/C, became the term for automated systems that were run by digitized control programs, ranging from the earliest punched paper tapes to today’s microcomputers.) As the specifications for military rotor blades grew more stringent, Parsons became increasingly concerned that his standard machining practices, which relied on templates and manual measurements, would not suffice.
In cooperation with the engineers and machinists in his shops, Parsons devised a new scheme for automatically controlling his cutting tools. Working from the blueprints for a rotor blade or other highprecision part, he specified the hundreds of closely spaced points that together would approximate the smooth curve of the contoured surface. He then recorded the location of each point (specified by Cartesian x- and y-coordinates) on a stack of punched cards and fed the cards into an automatic card reader. This, in turn, instructed the cutting tool to move from point to point.
Parsons’s system seemed to be an immediate success. It could produce parts to a tolerance of ±.001 inch, nine times better than the previous standard. The Air Force signed a contract for further development, and Parsons discussed a joint venture with Thomas Watson of IBM. The virtue of Parsons’s system, Noble suggests, was that it grew from an appreciation of the complications and realities of the shop floor:
For all his high-powered salesmanship, [Parsons] was motivated primarily by concrete problems in manufacturing and his sights remained fixed, not on military requirements alone or on technological advance for its own sake, but upon the real, changing, needs of industry.
Parsons enjoyed this initial victory in the late 1940s. From that point on, the irrepressible Parsons pops up in Noble’s story again and again, but his position becomes increasingly pathetic. Though a born tinkerer and promoter, he never made it big with his inventions. The reason, according to Noble, was that his ties to the shop floor placed him on the wrong side of military–industrial history.
The winning side of history was exemplified by Parsons’s chief competitors, the engineers at MIT. During the war, MIT contributed much to the military, and it received much in return. Its entrepreneurial engineers were just learning to look to the government for support. In the development of numerically controlled machine tools, the scientists saw a perfect opportunity for collaboration with the state.
To the military, collaboration with the scientific establishment was essential if the ever-higher performance standards it set in the late Forties and early Fifties were to be met. As its weapons grew more complex, so did the required manufacturing techniques. Instead of machines automatically controlled in three axes of motion, it now was asking for five-axis control.3 The MIT scientists, for their part, were seeking a mission and a sponsor for their “Whirlwind” project, an extremely costly and ambitious computer center. Whirlwind had originally been commissioned by the Office of Naval Research, but by the late Forties naval officials were denouncing it as “grossly over-complicated technically” and “notable for the lavishness of its staff and building,” and were preparing to cancel their contracts.
The Parsons numerical-control project, in which MIT was supposed to be a minor technical collaborator, held out new hope for Whirlwind. If the military’s design specifications became sufficiently recondite and mathematical, then a powerful computer, like Whirlwind, would become indispensable. There would be no other way to work out the control programs. Whereas Parsons, Noble says, “was primarily concerned about efficient and economical production and was seeking technical solutions to practical metal-working problems created by new aircraft design,” the MIT scientists
were concerned about furthering their research, advancing their professional careers as technical frontrunners, and in particular, developing the state of the art in the design and application of computer-based electronic control systems.
It was no contest. The Whirlwind team, notably Jay W. Forrester (who became more widely known twenty years later for producing the computer projections in the Club of Rome’s “Limits to Growth” report), quickly bested Parsons at every aspect of the government contracting game. Whirlwind grew in complexity and ability; Parsons bounced from one short-term project to another. Several of the MIT scientists formed a spinoff company, called Ultrasonic, to commercialize the innovations that Parsons had pioneered and the government had paid for. By the time Scientific American published a special issue on automatic control in 1952, Ultrasonic had moved far enough ahead with its plans to run its full-page ad opposite the table of contents. As Noble dryly notes, John Parsons’s name did not even appear in the issue.
MIT’s victory had several lasting effects, Noble says, but the most important was establishing automatic control as a purely mathematical discipline, rather than a practical one. John Parsons, a man risking his own money, built into his systems only the complications he was sure would pay off. Spending his working days with the men who would run his machines, he gave at least some thought to harnessing their skills and intelligence. But when writing the controlling programs became a branch of higher science, it could hardly be done on the shop floor. According to the MIT model, whose dispersal through civilian industry Noble exhaustively traces, there must be a clear division of labor along class lines. The men with clean hands would write the programs, and the men with dirty hands would switch the machines on and off. It was necessarily so: what could a machinist know about computer languages or differential equations?
Noble means to prove several points about industrial automation, and he is utterly convincing on two of them. The first is that, without such unrelenting pressure and support from the military, numerical control would not have enjoyed even the limited success it has now achieved. As long as the military was paying, scientists did not care how much the computers needed to design control programs cost. As long as they were making sales to the military, contractors did not care whether five-axis control systems made practical sense. Military specifications required the new systems, and military contracts covered the bill.
It was not until the mid-Seventies, when microcomputers drove the cost of computational power down, that numerically controlled machines qualified as even a partial market success, willingly paid for by firms that were investing their own money. Until that time, according to an analyst, Gerhard Widl, whom Noble quotes, it was used “mainly for military or aerospace products, where money seems not to be the limiting factor.”
Even for its subsidized users, numerical control could be a source of ceaseless headaches. Programs designed to prevent the operator from “tampering” with taped settings also prevented him from correcting the inevitable bugs and errors. Cutting tools wore out, tables got slightly out of alignment—and any of these imperfections could render an automatically controlled tool worse than useless. “One thing N/C can do quickly, efficiently, and automatically, one operator wryly observed, is produce scrap.”
Noble’s other contention, which in fact becomes the central argument of his book, is that social circumstances have determined the use of technology, rather than the other way around.
Although Noble quotes several of Lewis Mumford’s skeptical verdicts on the machine age, the guiding spirit of his book is not Mumford but Harry Braverman, whose Labor and Monopoly Capital was published almost ten years ago. In that book, Braverman offered the fullest and most eloquent contemporary exposition of the Marxist concept of “deskilling.” As technology advanced, he argued, it was used to displace labor selectively, with the most highly skilled, most independent, and best-paid workers the first to go. By deliberately leaching skill out of jobs, capitalists were said to use technology as a tool of class warfare against workers. Wage rates can be broken, since anyone can be hired for the mindless work of tending a machine. Control of the manufacturing process returns to the company office from the shop floor.
Noble’s application of the deskilling argument is most impressive when he demostrates that numerically controlled machines, which place power in the hands of scientists and managers, were far from a technological inevitability. In a long chapter called “The Road Not Taken,” he carefully describes another approach to automation, which would have been at least as “sweet” technically as the Whirlwind project at MIT.
The “record-playback” technique, known as R/P, had the same ultimate purpose as numerical control, namely, to permit firms to operate machines automatically and thereby reduce their reliance on skilled labor. The difference lay in the means by which controlled programs were prepared. The R/P control tapes were produced not by engineers sitting at their computer terminals but by skilled machinists, who “taught” the machines new jobs. Their movements were recorded and then played back to control other machines. Noble looks on this procedure more approvingly than on N/C because it acknowledged the workers’ skills:
With N/C, machinist skills were devalued and viewed as little more than a series of straightforward operations…. [But] R/P programming was, as one proponent described it, programming by doing…. The program, therefore, was a record not only of the machine (and stylus) motions but also of the machinist’s intelligence, skill, tacit knowledge, and judgment, which were embodied in those motions. Rather than viewing the possibility of human intervention cynically, as merely the chance for “human error,” this approach viewed it positively, as the opportunity for human judgment, skill, and creativity.
It is hard to doubt Noble’s argument that R/P was technically feasible. He offers case studies of its application at the General Electric factory in Schenectady, where Kurt Vonnegut had worked, and the Ford Motor Company’s River Rouge plant. Noble is not able to prove with similar crispness that R/P was abandoned for the reason he claims: because it threatened the managers’ class privileges. Such evidence is, by the nature of things, hard to come by, and Noble is left relying on suppositions and summing-up assertions, such as the following, about GE:
Given the emergent obsession with total, automatic control, a system such as this, which relied so heavily upon human skill in the preparation of tapes, was deemed obsolete upon arrival.
Similarly, Noble ends his persuasive and detailed chapter on GE’s Lynn, Massachusetts, factory with a free-swinging assertion. During the late Sixties, GE brought numerically controlled machines to Lynn under the “Pilot Program,” which allowed machinists to retain significant autonomy and control over the pace of production. The program seemed to be paying its way, financially as well as in factory morale. But executives in GE’s headquarters killed it, because of, Noble says, “a consideration far more fundamental than that of profitable production, namely the preservation of class power.”
Taken at face value, Noble’s laborious defense of record-playback is curious, to say the least. Perhaps R/P technique might have preserved the dignity of the machinists whose motions were set down on tape, but what good would it have done the many other machinists who would lose their jobs? Kurt Vonnegut did not need the specter of numerical control to envision what automation could do to the likes of Rudy Hertz. Noble, who has spent much of his life thinking about machine tools, has obviously anticipated such an objection, and he includes mild rebuttals in his book. His main response is that even after automation comes, R/P vests residual power in the machinists who remain on the job, however few they may be.
This argument has a pro forma quality because it is clear that for Noble’s purposes it does not much matter whether R/P would be significantly better for the workers. He is really interested in proving that the two approaches to automation, numerical control and record-playback, were roughly equal in technical merit. Having established no more than that, he can use the industry’s preference for N/C as evidence for his most important belief: that automation is undertaken not because “efficiency” demands it or the march of science makes it possible, but because it is a tool in the ceaseless war of capital against labor. If R/P would also deprive workers of their skills, so much the better for Noble’s argument. What better way to highlight the “contradictions” (a word Noble permits himself with increasing frequency as he builds toward his conclusion) of modern capitalism?
Unlike Harry Braverman, let alone Karl Marx, Noble has not made a frontal, systematic attack on the contradictions of capitalism. Having devoted nearly a hundred thousand words to a scrupulous history of numerical control, he merely points us toward his conclusion. He does not even directly condemn technology itself; in fact, he is at great pains to distinguish between the morally and socially neutral inventions and the militarists and industrialists who misapply them. But only the most obtuse reader can miss the melody behind the words: Noble is deeply apprehensive about industrial technology as it has been applied in the past and is likely to be used in the future. The last words in the book’s last chapter, which is called “Another Look at Progress,” are these:
the compulsion to automate (and to dominate)…continues apace (and resistance grows). As a result, we see, not the revitalization of the nation’s industrial base but its further erosion;…not the replenishing of irreplaceable human skills but their final disappearance; not the greater wealth of the nation but its steady impoverishment;…not the hopeful hymns of progress but the somber sounds of despair, and disquiet.
It is too early to be sure who will profit from the coming round of automation and who will be made to suffer; perhaps Noble’s warning of despair and disquiet will be borne out. But there are several reasons for one to hesitate before taking Forces of Production as a parable for the future of American industry.
How representative can the history of numerical control be, when it has affected so unusual an industry—and has made so small a dent even there? Noble does not dwell upon the practical difficulties that N/C users have encountered, but he never makes it clear what H. Allen Hunt and Timothy L. Hunt of the Upjohn Institute for Employment Research point out in their book: that a generation after Whirlwind and five-axis control, only between 3 and 4 percent of all metalcutting machine tools are numerically controlled. The Hunts study and, even more forcefully, the Office of Technology Assessment’s report concur with Noble that automation can often be used to build stress, monotony, and dehumanization into jobs. But both studies also emphasize that master machinists are hardly typical of workers who lose their jobs to automation. Few who are displaced are skilled craftsmen, whose years of experience are suddenly rendered obsolete, according to the Hunts and the OTA. Most are assembly-line welders or painters, whose jobs are already stressful, monotonous, and deskilled.
In fact, both reports say that the effect of automation on unskilled workers is far greater than its threat to machinists or other craftsmen. Both reports contend that automation is a trivial source of unemployment compared to swings in overall economic demand. For example, the Hunts estimate that by 1990, the use of industrial robots might displace between 4 and 6 percent of all automobile workers (and less than 1 percent of the entire American work force). By contrast, the auto industry lost one third of its production workers between 1978 and 1982 for reasons that had nothing to do with automation—except, perhaps, that the auto-makers had been too slow to adopt it.
Even in the machine-tool industry, there is another side to the malign advance of automation. Noble’s history runs from the late 1940s through the 1970s. As the OTA points out, those same years brought rising prosperity not simply to American workers but also to machinists. Between 1947 and 1980, average weekly earning for metal-working machinists rose from $58.69 to $346.83, and average hourly earnings rose from $1.38 to $8.18. The conjunction of rising wages and more advanced technology was no accident, the OTA contends. Each factor enhanced the other.
This argument, too, David Noble has anticipated, and he alludes to it in an offhand remark that does much to clarify the arguments about industrial automation. Noble refers to a Fortune article published in 1946 which examined the possibility that postwar automation would create chronic unemployment. The authors concluded that, on the contrary, automation would create many more jobs than it would destroy. Noble says of the article:
in the time-honored fashion of apologists of progress since the first Industrial Revolution, the authors remained sanguine about the possibilities, and portrayed the displacement of labor from their traditional means of livelihood as the liberation of humanity from unwanted and necessary toil.
The crucial words are “traditional means of livelihood.” In a commercial society built on constant adaptation to technical and economic change, it is hard to know precisely what they signify. For two hundred years, Americans have left their traditional places on the farms to fill the cities, the factories, and—yes—the machine shops, where they developed the skills that Noble properly esteems. Some of these episodes have been brutal. The young women of nineteenth-century New England, who left their rural homes to toil in the mills of Lawrence and Lowell, and the black men and women who left the land in Alabama and Georgia for factory jobs up north, suffered at least as grievously as a deskilled machinist.
Where the suffering was avoidable—as, most obviously, it was for the blacks—the state should have done more to prevent it. (Had the US Department of Agriculture been less a haven for racists during the Thirties and Forties, more black farmers could have stayed where they wanted to be, on the land.) But should the entire process of adaptation be challenged? That further technological revolutions will deskill and impoverish on a large scale remains, in any case, unproven, even after this book. That they will upset “traditional” life is inevitable; that is the cost to bear in mind as the next revolution begins.
September 27, 1984
The most important and elementary machining concept, which Noble alludes to hundreds of times but unfortunately never explains, is “speeds and feeds.” Speeds are the rates at which the cutting tool is moving—or, in the case of a lathe, the rate at which the work to be cut is rotating against the tool. Feeds are the rates at which the tool makes its cuts across or into the piece of work. His control over “speeds and feeds” and other variables, so as to produce a finished piece to the specified tolerances, is the essence of the machinist’s skill. The Office of Technology Assessment’s book Computerized Manufacturing Automation contains a very useful illustrated primer on machine-tool operations. ↩
Military technology is not really Noble’s subject in this book, but he implies that the military’s demands for closer and closer tolerances arose from its undisciplined taste for complex technological solutions, even when they made no strictly military sense. For example, he says that aircraft specifications, which required new, automated machining techniques, ↩
This improbable-sounding concept, according to Noble, involves positioning a tool along three axes, and rotating or tilting along two others. ↩