Automation Doesn’t Have to Mean Unemployment and Misery
Automation won’t necessarily lead to either mass unemployment or a utopian workless future. Power and politics impact how automation affects work — which means that automation can create dignified jobs through class struggle.
The threat that automation is in the process of destroying large numbers of jobs has been much discussed recently. Sober think tanks such as the Brookings Institution and the McKinsey Global Institute have forecast that automation will eliminate tens of millions of US jobs in the next few decades. The Atlantic has dedicated eight thousand words to a feature on “A World Without Work.”
In his new book, Labor’s End: How the Promise of Automation Degraded Work, labor historian Jason Resnikoff reminds us that we have been here before. From the 1940s to the 1970s, much of the US intelligentsia was transfixed by the notion that technology would in short order largely eliminate manual work. Some saw this as a welcome development that would eradicate drudgery and usher in a post-scarcity age of plenty, others as a looming threat requiring bold action to safeguard the well-being of masses of workers. But all, with few exceptions, saw this change as inevitable.
Five-plus decades later, our country still has an abundance of drudgery, along with scarcity for many, plenty for some, and vast excess for a few. And we still have a great deal of manual labor, though less of it involves metal-banging and more involves serving customers or providing care.
In Labor’s End, Resnikoff argues that we should not be surprised. “Automation,” he maintains, placing the word in quotes throughout the book to underline his point, never consisted of actual labor-saving technological transformations in production processes. Instead, it was an ideology that served to obscure the grim reality of how businesses were reshaping workplaces. Onstage was a vision of an efficiency revolution that would sharply reduce or even dispense with the need for human labor in factories, offices, and homes. Backstage were speedup, de-skilling, and, in many cases, heightened occupational hazards, along with other attacks on workers’ bargaining power via relocation of work to lower-wage areas. The avatars of automation sold the idea that triumphing over the constraints of nature was the way to end degraded work; automation’s critics accepted this premise but insisted on a fairer division of the remaining jobs and of the bounteous output that would come with technological advances. Both thus directed attention away from the ways that power and politics continue to determine the degradation or dignity of work. Is today’s automation obsession repeating this misdirection of attention? I will return to that question at the end of this review.
Resnikoff traces the word “automation” back to a 1946 utterance of D. S. Harder, Ford’s vice president of production. As with many terms in good currency, others claimed to have invented it, notably electronics innovator John Diebold in his 1952 book, Automation: The Advent of the Automatic Factory. Much of the narrative in Labor’s End documents the extent to which a wildly diverse group of influential people embraced the concept and some version of the vision it implied — sociologist Daniel Bell, to be sure, but also United Auto Workers president Walter Reuther and computer scientist Norbert Wiener; presidents John F. Kennedy and Lyndon B. Johnson but also radical intellectuals like Herbert Marcuse. Resnikoff conveys this cacophony of voices — along with a smaller number of dissenting ones — by alternating between case studies of the implementation of “automation” (in auto manufacturing, coal mining, meatpacking, office work, and housework) and intellectual debates (“Is freedom compatible with industrial capitalism?,” “Is the working class still the agent of industrial change?”) that were recast by the conviction that manual labor was rapidly becoming a thing of the past.
Labor’s End makes four particularly noteworthy contributions to our understanding of work in the United States in the late twentieth century, each of which sheds light in turn on work today. First, Resnikoff deftly depicts the bait and switch US industrialists carried out, promising a technology-enhanced lightening of laborers’ burdens but actually delivering speedup. This is not a novel insight; Harry Braverman’s debate-shifting 1974 book, Labor and Monopoly Capital, made the point powerfully, and Resnikoff cites varied work by like-minded labor historians to construct his case studies. But this book does a great job of contrasting the lofty rhetoric of executives like Diebold with the firsthand testimony of factory workers pushed to the brink of exhaustion and clerical workers struggling with anxiety, boredom, or both.
Second, the book makes provocative connections between the late-twentieth-century automation discussion and long-standing debates about work throughout the history of the United States and, for that matter, the world. Thomas Jefferson and Alexander Hamilton famously debated the best way for the newly independent United States of America to combine economic prosperity with democracy and freedom. Jefferson, somewhat ironically for a large landowner and slaveholder, argued that liberty could only be safeguarded in a “yeoman republic” consisting of small farmers and independent producers of roughly equal means, with a limited government guided by deliberation among these yeomen. Hamilton insisted that economic progress depended on industrialization and thus on a larger government to both nurture and regulate industry, whereas Jefferson feared industrial development would lead to concentrations of economic power and replication of Britain’s “satanic mills,” creating a miserable, ignorant, vice-afflicted working class. Karl Marx, of course, saw the resolution of this conflict via the working class’s collective seizure of control over the means of production — and in the militant US labor struggles of the 1930s and 1940s, large sections of the working class supported some version of Marx’s prescription for building worker power. The automation ideology, Resnikoff posits, suggested instead that the resolution was putting to use the newly available technologies that would eliminate all those immiserated manufacturing jobs and transition workers to clean and pleasant white-collar work.
But Resnikoff maintains that automation promised to resolve an even longer-standing philosophical dilemma. Aristotle, observing that humans must engage in tedious and difficult labor to survive, concluded (mirroring the structure of ancient Greek society, divided into citizens and slaves) that those of aristocratic nature, fit to make important decisions and think creative thoughts, should be spared unseemly toil, whereas their inferiors, best suited for arduous labor, should specialize accordingly. The automation discourse suggests, again, that we can dispense with the need for Morlocks and all be Eloi. Whereas corporate heads like Diebold airily implied that this would happen more or less spontaneously, worrywarts like Reuther and Willard Wirtz, Kennedy’s labor secretary, insisted that a successful transition would require robust national policies for retraining and redistribution of the productivity dividend. Meanwhile, traditionalist conservatives remained loyal to Aristotle’s schema, holding that the path to the good society lay not in transcending labor but in returning to time-tested values and hierarchies abandoned in the rush to modernity.
A third gift of Labor’s End is showing that much of the 1960s New Left bought the story that automation was leading rapidly and inexorably to the vaporization of industrial work. Resnikoff’s evidence includes the writings of the neo-Marxist Marcuse, but also eco-anarchist Murray Bookchin and African American socialist worker-intellectual Carl Boggs. Even more compelling is the image of Students for a Democratic Society leaders Todd Gitlin and Tom Hayden mobilizing their members to organize masses of displaced workers, only to conclude that, in Gitlin’s frustrated words, “We failed, and are still failing, to make a case for the quantitative impact of cybernation.”
Finally, Resnikoff draws intriguing parallels between left views of the future of work and feminist critiques of the drudgery of housework. In The Feminine Mystique, liberal feminist Betty Friedan asserted that, while homemaking had once been meaningful work, automation of the home had usurped most of those tasks, leaving housewives trapped in vacuous make-work. Radical feminist theorist Shulamith Firestone, on the other hand, saw automation as women’s potential savior. She advocated a revolution that would “redistribute drudgery equitably, but eventually eliminate it altogether” through “cybernation” of household tasks, including childbearing itself. In other words, both authors presumed that automation could end the need for domestic drudgery. Of course, another ideological parallel, not noted by Resnikoff, is that once again cultural and religious conservatives insisted that women’s fulfillment lies not in automating domestic drudgery to free them for more stimulating activities but in embracing their “natural” role.
Conservatives were not the only ones who dissented. One of the most fascinating contentions of Labor’s End — more of a sketch than a fully developed argument — is that key black civil rights leaders, rather than being distracted by the prospect that automation would eradicate bad jobs, demanded that American society invest even the humblest of tasks with social value and economic rewards. Thus, Martin Luther King Jr supported striking Memphis sanitation workers in 1968 by maintaining that “all labor has dignity.” Around the same time, National Welfare Rights Organization leader Johnnie Tillmon urged a presidential proclamation that “women’s work is real work” and said that mothers should receive a living wage for “doing the work we are already doing — child raising and housekeeping.”
Resnikoff’s book is well argued and bolstered by copious evidence, but he does sometimes overreach. Perhaps the most serious of the overreaches is his implication that, at least for the most part, managers installed the new equipment billed as “automation” simply to disrupt old ways of producing and introduce speedup and that increasing efficiency was never on the agenda. He does concede at a couple of points that true productivity increases occurred — “Yes, the introduction of machinery could reduce the amount of human labor necessary to produce goods,” he acknowledges in his conclusion — but these concessions can be hard to spot in the barrage of arguments to the contrary.
Yet productivity increases are real and widespread. Consider telecommunications. A thought experiment suggests that if each telephone (or Skype or Zoom) call required one or more operators to manually make connections on a switchboard, either we would be making many fewer calls today or a huge proportion of the world’s workforce would have to be employed as phone operators. But we don’t have to rely on a thought experiment when my own experience will suffice. In the last several decades, I have gone from working on a semiautomated phone switchboard (pressing buttons to direct calls) in the 1980s, to visiting call centers with automated call routing in the 1990s, to using an autodialer and Skype in the 2000s, to Zooming in the 2010s.
The same is true for manufacturing. As a simple measure of manufacturing-wide productivity, we can look at inflation-adjusted value added per manufacturing worker over time, combining data from the Bureau of Labor Statistics and the Bureau of Economic Analysis. By my calculations, this measure of output per manufacturing worker more than doubled between 1947 and 1974, and more than doubled again between 1974 and 1997. Some of this is due to shifts in the mix of manufacturing (less apparel manufacturing and more microchip manufacturing over time, as the global division of labor changed), and some may be due to the speedup that Resnikoff emphasizes, but the majority is surely reflecting more productive technologies. In auto assembly, Resnikoff’s favorite example, robots have taken on growing amounts of work previously done by humans since the 1980s. Similar patterns hold in steelmaking and coal mining, in which entirely new technologies — minimills and open pit mines, respectively — have led to marked increases in efficiency (though in each case only for lower grades of product).
How can this be consistent with the evidence Resnikoff assembles showing that when new equipment was put in, worker head counts increased or remained unchanged? There are two parts to the answer. The first builds on economists’ observation that installing more efficient equipment can have both “substitution” and “output” effects on the amount of labor employed. Substitution effects refer to substituting machines for labor, hence resulting in fewer workers employed. But output effects take into account that when equipment lowers the cost of production, product prices tend to fall, and consumers buy more, boosting the number employed. If output effects exceed substitution effects, adoption of labor-replacing machinery leads to an employment increase. The second part of the answer is that new technologies always involve a learning curve. Early in that process, the new systems fail frequently, and machines break down, but over time, workers and managers learn how to make the technology work efficiently (or else management usually abandons it).
Though Resnikoff’s dismissal of technology-based productivity is off the mark, his point that business used new machinery to shake up job descriptions and introduce speedup is right on target. Corporations have long applied a sort of “shock doctrine,” using mechanization to justify the rewriting of work rules. That reality eventually sparked a surge of worker discontent that leads Resnikoff to end his history in the early-to-mid 1970s. In those years, rank-and-file workers fed up with the speedup, degradation, and de-skilling that accompanied automation showed their unhappiness through strike waves, sabotage, and widespread alienation. Between the obvious fact that factory work (along with clerical work and, for that matter, housework) was not by any means disappearing and the now-evident fact that automation was not leading to improved jobs and happier workers, the mystique of automation was shattered.
But it was not to be the automation ideology’s last act. Resnikoff describes our current moment as the third wave of automation discourse (the second ran from the late 1980s to the early 1990s). I research how technological change is reshaping retail jobs, and these days, the trade press is full of quotes that echo those in Labor’s End, like this one (from RIS [Retail Info Systems] News: “Most retailers adopting automation face the risk of pushback from employees. . . . Companies need to provide reassurance and communicate clearly about what they’re offering — an exciting new journey that will require employee engagement.” Most of the current discussion of technology and work adopts the same flawed assumptions that Resnikoff challenges in his book: technological change develops autonomously and spreads inexorably; technology and its applications to work are apolitical.
There is, however, one important difference between the mid-twentieth-century automation discourse and that of today: the earlier optimism has been replaced by widespread apprehension. Rather than “The robots are coming, and life will soon be better for all of us,” the predominant refrain today is, “The robots are coming, and they’ll bring huge disruption — so we need to figure out how to adjust to it.” The leading proposals for adjustment involve massive retraining (advocated by orthodox economists and the liberal policymakers who heed them) or attacks on immigrants or foreign competitors who are allegedly stealing “our” jobs (the policy menu of the Trumpist right); both are deeply flawed.
The problem with both stories about the robots, as Resnikoff warns, is that they assume the robots are marching forward regardless of what we do, whereas in fact human actors are the ones controlling what technologies get developed, which of them are used, and how. Seeing that clearly allows us to break away from an impoverished set of policy choices and begin articulating a progressive view of how to reshape work. King was right to insist that we make every job a dignified job. We can do that by raising the floor of wages and working conditions, by increasing workers’ voice at work via unions and other organizations, and by redefining jobs to draw on broad skills and encompass varied tasks. In Germany, for instance, most retail workers have been trained through a two-year apprenticeship, are qualified to do almost any task in a store, and benefit from a nationwide contract negotiated between the retail union and the industry association — a far cry from the situation in the United States. Tillmon was justified in demanding wages for socially valuable but unpaid care work. Indeed, we should extend this demand to other kinds of socially valuable work, such as community-building, as economist Nina Banks and others have argued. And we should also emulate King in advocating an adequate guaranteed annual income, taking steps to separate income from its dependence on either work or property and giving workers a fallback that will allow them to turn down the worst jobs and insist on something better.
The history recounted in Labor’s End helps arm us to counter fallacious reasoning about automation and advocate for shifting the workplace toward greater worker power, dignity, and prosperity. Resnikoff’s probing analysis directs our gaze away from the “shiny objects” of new technology and redirects it to where it belongs — on workers.