Black Lives Matter. We stand in solidarity with those affected by generations of structural violence. You can help »

The Brooklyn Rail

JUNE 2021

All Issues
JUNE 2021 Issue
Field Notes

The Future of Automation

Aaron Benanav
Automation and the Future of Work
(Verso, 2020)

Jason E. Smith
Smart Machines and Service Work
(Reaktion, 2020)

Probably no one needs as much convincing regarding the future of automation and its impact on employment as provided by these two timely and important books—Jason E. Smith’s Smart Machines and Service Work: Automation in an Age of Stagnation and Aaron Benanav’s Automation and the Future of Work. Each serves as a calming antidote to the often hyperbolic pronouncements emanating from popular and scholarly accounts.

Remarkable about both books is the degree to which they cover the same terrain and use similar arguments to debunk a cult of automation that has excited nouveau-billionaires like Elon Musk while frightening others like Andrew Yang. That these books parallel one another so closely suggests that they represent a consensus about the economy within the alternative and non-parliamentary lefts. As important as the discussion of automation is, however, it is where these two books deviate from one another, both in their methodological approaches and in the theoretical traditions upon which they draw, that the discussion really gets interesting. But first, to automation.

That capitalism is a cost-reducing and labor-saving mode of production captures what most everyone agrees on, regardless of their theoretical and political orientations. Karl Marx made this the centerpoint of his analysis, but this insight into the functioning of a competitive economy is shared by even the most woolly-eyed and naïve of economists. To make more with less, so it’s said, propels economic growth. For both the individual entrepreneur and the modern corporation, cutting costs in order to increase productivity and profit is the logic behind their actions. The tendency to automate, as both our authors point out, is nothing new, but continues long-standing processes replacing real human beings with machines and substituting expensive machines with either less expensive or more productive versions. Automation has always been essential to economic development.

Implicit in both books is the degree to which manufacturing and the production of commodities—as opposed to the assembly, warehousing, and distribution of components and finished goods alike—are already fully mechanized. Industrial mechanization, viewed in terms of the quantity of goods produced with relatively small workforces, is a defining feature of the modern industrial system and a key factor that limits future possibilities for fully-automated processes. Handicraft production exists only on the peripheries of the global system.

If neither Smith nor Benanav is overly concerned with the history of technology, it’s because they have chosen alternative paths to get to similar endpoints. Further automation is restricted by mechanical, profit, competitive, and consumption limits dictated by the peculiar economic order under which we live. Each of these presents insurmountable barriers, and each in turn can be used to understand the limits that confront the entire system of production.

Because automation has received so much attention in the media, the public tends to be aware of exceptional situations rather than those that constitute the norm. Industrial production mostly takes place behind closed doors, that is, in windowless or near-windowless buildings located in industrial parks or zoned districts that are geographically distant from middle- and upper-class residential districts and shopping areas. If you don’t work in a factory, you probably don’t live near one either. Despite our dependence on industrial production, few people are knowledgeable about the working conditions, design, or employee qualifications in manufacturing enterprises.

In “boutique” areas such as the steel manufacturing of the last several decades, highly-skilled craftspeople produce goods in greater quantities and complexity than anything imaginable a half century ago. The “heavy” industry of yesteryear that relied on muscle-power and masculinity has been replaced by machines that are monitored, repaired, and retooled by skilled operatives—frequently college-educated—adept at the use of computers and mechanized devices. A cursory look at Wikipedia, for instance, reveals that in 2014 in the United States, 218,000 workers in iron and steel mills and foundries produced 29 million metric tons of pig iron and another 88 million tons of steel. This equals over one million pounds of metal per employee.

In terms of production, two well-known exceptions involve areas of great importance to consumers. The meat-packing and textile industries are both notorious for their low wages and mistreatment of employees by means of brutalizing working conditions. Employees are kept in a pitiful state, unable to support themselves decently and subjected to dangerous and mind-numbing work. Even though, as both Smith and Benanav explain, capitalists have a motive for replacing low-waged living labor with machines when the tasks performed are simple and repetitive—precisely the type of tasks that machines perform readily—low wages also undercut economic development whenever cost savings through the introduction of machines are not great enough to warrant the expense, especially—as Benanav emphasizes—in commodity areas that seemingly are already at a saturation point. Despite the existence of tens of thousands of prototypes, models, and designs for labor-saving machines and procedures—one need only to visit an industrial tradeshow to be convinced of this—the lack of sufficient financial incentive mitigates against the further development of technology. This dilemma gets to the heart of the two books under review.

The automation that the public hears about today centers on the assembly, warehousing, and distribution of goods, on infrastructural projects, and on the attendant accounting processes, but not generally on manufacturing. Large-scale infrastructure developments receive extensive media coverage. Projects such as bridges and skyscrapers, for instance, require few workers to use the heavy equipment—often one operator per machine—needed to build the metal and concrete frameworks upon which roadways, apartment houses, and office buildings are erected. Much of the power grid of electrical generating stations, nuclear plants, and natural gas storage areas and piping systems are similar—small workforces monitor complex sites that are about as fully mechanized as is technically possible. Container ships and freight railroads are familiar examples in which minimal staffing is needed to transport enormous quantities of goods. The tanker recently run aground in the Suez Canal, for instance, was loaded with nearly 20,000 metal shipping containers, each weighing some 20 metric tons on average; yet the tanker had a staff of only 25. Recycling and water purification facilities are still other areas that involve substantial investments, complicated machinery and systems, and minimal numbers of highly-skilled employees.

Both books devote their opening pages to disentangling fantasy from reality with regard to automation, the current version of which claims as actually possible the full replacement and displacement of living labor; in other words, an automation that presupposes self-correcting and possibly self-generating mechanical and electronic systems that require little or no human attention. As Benanav notes, it’s easy to “mistake the technical feasibility of automation…for its economic viability.” At one extreme, Elon Musk promises a life of perpetual leisure with luxury automobiles and cruises to the moon, while at the other Andrew Yang flashes the nightmare of mass unemployment and guaranteed incomes pegged at a level below subsistence.

More realistic discussions of automation focus on its ability to transform entire areas of society. The companies that have risen to power over the past several decades are prime illustrations. Their skill at creating new consumer needs and preferences has had a profound effect, although often success has meant combining automation with huge cost-saving economies of scale. To do so, however, these companies had to first cannibalize existing businesses. This was the road to dominance in case after case. Home Depot undercut local hardware stores, lumberyards, plant nurseries, appliance outlets, home decorating stores, and more. Walmart did the same to small retail clothing establishments, mid-sized department stores, and other businesses also cannibalized by Home Depot. Uber and Lyft decimated taxi and limousine services. Apple devoured considerable portions of the communications industry. Amazon’s destruction of the pre-existing economy is a story often told.

Even though these companies relied on the latest in communication platforms and mechanized warehousing and supply chain systems, underlying their success was and is a heavy reliance on sweated labor and a massive, long-term, and highly speculative investment—drawn from the vast amounts of free-floating capital unable to find profitable outlets. These are the preconditions underlying the new economy on which the future is based. The United States already had the world’s most developed retail sector before these companies pushed it aside.

Not until later in their development, after these companies have grown into near- or full-monopolies in their respective economic sectors, do they begin to produce profits, and then not always since a rise in stock prices can be sufficient to keep attracting investment capital. This ability to garner a larger share of total social wealth comes at the expense of other sectors of the economy, for whom less remains. While a limited number of companies dazzle consumers and investors alike, all other business entities face intensified competition and a paucity of profit that appears to have no specific reason except the state of the economy at large.

This pattern—of plundering the economy in order to make room for new technologies and types of business organization—is more extreme than anything that took place previously, when cutting edge technologies and businesses were known primarily for their creation of new economic domains and the further development of relatively underdeveloped aspects of the economy. Intense competition may have existed among the upstarts, but displacing the pre-existing economy was not a major part of their business plans, given how rudimentary the past was in comparison with the future.

The economic deadend represented by automation today—and capitalism in general—is the theme to which both Smith and Benanav dedicate themselves. Their books paint a bleak picture of economic mal- and mis-functioning, with rates of investment and growth tending toward zero, stagnant wages the norm for a half century already, “zombie” companies kept afloat through debt financed at low interest rates, and either thin or erratic profit margins everywhere. Benanav concentrates on the dilemmas faced in manufacturing, while Smith focuses on services. Another contrast is Benanav’s reliance on mainstream economics, while Smith turns to the critique of political economy. Along the way we are provided with quick benchmarks in order to orient ourselves, a quality most necessary in books that focus on the ideology of automation and the economic theory that undercuts it, and less so on history and contemporary examples. Either way, there’s lots of overlap in these two books; after all, it is the same reality that each explicates.

Automation and the Future of Work

Benanav’s Automation and the Future of Work works best as an economic history of the last half century, when a unique combination of factors entailed a growth in government expenditures simultaneous with a hefty effort to privatize government services. This, of course, summarizes the odd confluence of circumstances known these days as neoliberalism. If Benanav’s is the most cogent of the two books, it’s because he hews closely to a singular economic interpretation.

At its center is industrial overcapacity, that is, the ability to produce goods in excess of actual demand. This tendency towards overproduction brings in its train a host of secondary phenomena—the focus of Benanav’s analysis—such as declining rates of investment, stagnant growth (secular stagnation), and inadequate profit rates, which in turn prompt intense efforts to cut costs, move production facilities to lower-wage and lesser-regulated regions of the world, and expand and streamline supply chain networks.

Most interesting are the relationships Benanav draws between rates of growth in three key economic dimensions, that of productivity, output, and employment. Through this lens, he presents a comprehensive overview of recent economic development and an analysis of the economic doldrums in which much of the world’s economy now finds itself. Benanav’s approach—through rates of growth rather than absolute numbers—offers a good means to understand highly complex relationships. Output, for instance, can only grow more rapidly than productivity when employment increases. This, he points out, was the situation that prevailed in the few brief decades following World War II, as an unprecedented prosperity produced ample profits and, at the same time, more jobs and higher wages.

Automation, on the other hand, promises to reverse these relationships, so that productivity outpaces actual production. Employment thus lags, a circumstance that creates a surplus of available employees and ample opportunities for employers to control costs by means of wage freezes, two-tiered wage systems, underemployment, fewer or no benefits, and other measures that come at the expense of the workforce. For Benanav, global overcapacity, not automation, accounts for the slowdown in growth and productivity that characterizes the recent era. Overcapacity, in this view, becomes the driver behind deindustrialization, whereby quite ferocious efforts are made to rekindle a level of economic success that has become all-too-elusive.

Benanav, though, wanders back-and-forth across the borders that distinguish economic history from economic theory. Overcapacity, while it describes the economic doldrums of the past half century, is also common to all phases of capitalist development, its periods of growth as well as periods of stagnation and outright decline. Without overcapacity and the overproduction that accompanies it, much of the rationale for both competition and commerce would be lost. The entire retail sector, to take one example, depends on overproduction as a condition of existence.

There are times, of course, when demand outruns production, but this occurs mostly because supply chains have been interrupted due to natural or political occurrences, not because of an inability to step up production. When it occurs, the pricing mechanisms applicable to supply and demand kick in, thus allowing industry time to adjust to an altered situation. One of capitalism’s genuine accomplishments over the past two centuries has been its ability to close the gap between surges in demand and the economy’s ability to scale up. The recent pandemic is an example of just this. Despite all the chaos and political bickering, and the terrible and needless suffering that ensued, the ability of industry to develop and test vaccines, and then produce them in tens of millions of doses, took place within an eighteen-month framework.

The theoretical question is why overcapacity and overproduction in some phases of capitalist evolution are spurs to further development, whereas in the recent period focused on by Benanav, they have become hindrances to be overcome. To note too is that economic growth, even sluggish growth, entails that both production and consumption take place at ever-higher levels, another of the phenomena that remain unintelligible if the focus remains fixed on overcapacity. Historically, this ramping up of the economy has been particularly noticeable following economic downturns, when both consumption and production emerge on unprecedented levels. If overcapacity is the root cause of economic sluggishness, how is it possible, then, for an ever-greater capacity to serve as a precondition for an economic resurgence?

Benanav’s politics are not always consonant with his analysis. The focus on overcapacity fits entirely within the economic analysis associated with John Maynard Keynes, in which a lack of demand, it is thought, can be counteracted with targeted government spending. Benanav has no faith in such solutions, given the largesse of governmental interventions over the past half century and the ongoing decline in business fortunes as measured by rates of output growth and productivity.

Were governments to push further in terms of an outright nationalization of industry, he notes, they would face disinvestment and capital flight. Short of a wholesale expropriation of business property ala Russia during the early twentieth century, however, business owners have historically accommodated a wide range of political systems, from democracy to fascism, with all shades of authoritarianism and social democracy in between. They may favor regimes that offer the most support and latitude in terms of contracts, subsidies, tax write-offs, minimal wages, and lax regulations, but this tends to be relative rather than absolute.

At times, the business world favors regulation because of the latter’s ability to neutralize cut-throat competition and help ensure oligopolistic operating conditions. Trade agreements are a case in point. Disinvestment, when motivated by political considerations rather than economic ones, comes at a cost, and consequently is rarely resorted to, despite the attention these sorts of occurrences receive.

Benanav’s discussion, in any event, presupposes a sharp divide separating government from business, an assumption that itself is worthy of critical attention since the same personnel tend to float between both worlds. Even when politicians hail from within popular movements, rather than from the business and legal worlds, they need to rely on these latter groups for their expertise. Policy and policy-making is common to them all.

Smart Machines and Service Work

Smith’s Smart Machines and Service Work is less tightly structured, and in another era might have been titled Towards a Theory of Machines and Work. The data-laden sections in the first few chapters are the most difficult, although they cover similar terrain to Benanav’s Automation and the Future of Work.

Particularly illuminating are the passages which provide descriptions of low-wage jobs that defy attempts at automation, for example, when the tasks performed are too local in nature, like neighborhood restaurants, or too complex, like lifting and bathing ill patients without bruising them—jobs “identified with unpredictable, highly intuitive decisions and activities that are nevertheless deemed ‘human’ or ‘natural’, instinctual or innate, even though they tend to be subtle, learned capacities cultivated within the context of private or family life rather than in school or at work.” Personal services represent economic areas somewhat immune to mechanization, and consequently serve as a drag on attempts to enhance the economy’s overall productivity and effectiveness.

Smith is best as an analyst of theory as demonstrated by his many excellent essays over the years. Early on in his book he turns to the misleading nature of economic statistics, itself an indication of the blinders with which analysts of the economy function. Within a specific branch of industry, for instance, productivity can be measured by comparing hours worked with output. An increase in the latter indicates enhanced productivity, either because the workforce has been sweated more intensively, other costs reduced (less waste, for instance), or because of newer and efficient technologies.

Creating a productivity index for an entire economy, however, proves elusive without first resorting to a universal equivalent (e.g. money) through which rates of productivity for specific products and services can be compared. These indices, however, must control for price fluctuations, and since rates of inflation and deflation vary from commodity to commodity and on a daily basis, this transforms the best methodologies into informed guesses. The price of crude oil is a well-known example, and what’s true for oil is true for every commodity for which oil and its byproducts, such as fuels and plastics, are components.

Along with productivity rates, business costs as well as business profits are also “best” estimates. To approach this same issue from another perspective: an automobile contains some 1,800 discrete components, which can be further broken down into 30,000 separate pieces. Calculating the exact cost is an impossibility, no matter how large or dynamic the spreadsheet. If costs aren’t accurately calculated, neither are profits. Smith, though, like everyone else, has no choice except to resort to these data to describe economic trends.

The discussion of data takes us squarely into the strongest sections of Smith’s book, where he uses key concepts from Marx’s Capital to explain the dilemmas within which automation and the economy at large remain trapped. The difficulty for economists to distinguish between the physical and value aspects of the economy, as in measures of productivity, gets to the heart of Marx’s brief formulation of the “organic composition of capital,” a key concept largely overlooked within the history of Marxian economics but which should have been crucial in deciphering the twentieth century.

The bifurcation between manufacturing- and service-oriented jobs is reframed by Smith in terms of the Marxian concept of unproductive work: employment that even when it produces surplus value, no longer contributes to the process of capital accumulation. Smith’s discussion is based on relevant passages from Marx’s Grundrisse, although just as helpful is the extensive, albeit also incomplete, discussion in his likewise posthumously published Theories of Surplus Value.

These distinctions are especially appropriate for the service sector, where small and mid-sized businesses continue to exist in large numbers. Small businesses especially have long been stuck in the revolving door of simple reproduction. At best, they are able to pay their staff, their owner-operators, and their expenses, but thoughts of expansion and renovation, as opposed to survival, are unrealistic. Smith analyzes these businesses in terms of the types of work performed, rather than the more customary approach which focuses on the size of the establishments that offer services as commodities.

From the historical sections of Marx’s Capital, Smith draws our attention to passages that describe the impact of mechanization on other non-mechanized sectors. For Smith, this points to “a core contradiction of the capitalist use of machinery: the very productivity of capitalist industry consigns a larger and larger portion of humanity to low-productivity, and often unproductive in Marx’s sense of the term, laboring activities.” This theme, shorn of its foundation in Marxian value theory, is important to Benanav as well. In Marx’s day, it was cheaper to hire destitute women to pull barges along inland canals than to purchase the machinery that could perform the same function.

These days, employment precarity accompanies futurist visions of a fully mechanized singularity in which machine knowledge (artificial intelligence) would have the ability not only to replicate and repair itself but also to self-generate new modes and levels of functioning. Smith notes instead the huge expansion of supervisory and managerial labor that has accompanied the intensification and marginalization of work at all levels of the socioeconomic continuum.

These insights, along with other discussions throughout Smart Machines and Service Work, lend it a freshness unusual for books that address such ponderous topics.

Politics Within Automation

In politics, Smith takes us to the abyss of revolutionary upheaval and social transformation by distinguishing three types of protest, each of which carries profound implications. The technical division of labor is focused on the production process and associated with the traditional labor movement, which with deindustrialization has found its realm of action severely restricted. The second arena of protest is newer in genesis and is centered on a social division of labor, in which teachers have played a prominent role in recent years. Here, “jobs remain largely invulnerable to both automation and offshoring…because the tasks performed do not admit replacement by even the most advanced technological innovations [and] because these are in-person services performed on site.”

A third, less easy-to-define arena of protest is typified by the Black Lives Matter movement in the United States and the Yellow Vests in France, the latter largely rural and suburban supply chain employees who are not clustered at work and who nonetheless conducted a long series of widely popular actions. Smith does no more than categorize these protests, in part to show that the ongoing fixation on the old labor movement of unions and political parties is out of touch with the tactics and centers of activity that have recently emerged.

Benanav, on the other hand, beckons to us from across the abyss. His discussion of the two realms of freedom and necessity is beautifully written and intrinsically interesting, but it shares with utopian thought in general the disconnect between the critique of contemporary conditions and the projection into the future of a possible alternative.

Missing in this framework is a discussion of how a revolutionary transformation might take place in the here and now. If massive protests began this evening, for instance, what might people do to further and solidify a new way of existence? Gather in large groups in front of city hall and other public buildings? Demand new elections and more responsive political representatives? Or perhaps, preferably, proceed to work, school, and government agencies and occupy their premises? Then what? How do these types of actions lead to a sustainable system that cannot be taken away by property owners and the thugs newly out of work and eager for even a precarious existence as hired guns? And how might these new systems of economic and social functioning be organized such that no single group can usurp the decision-making abilities that determine everyone else’s fate?

Benanav tends to project the present into the future, at times advocating the “abolition of private property,” but at other junctions speaking of “industries partially socialized.” Despite his skepticism about the UBI idea, “a basic income could form one part of a larger social project aiming at human freedom.” Elsewhere he mentions an “end of scarcity” that would free people to enter “federations for building spaceships,” among other activities. Yet, it is hard to think about federations or spaceships without imaging a future that still includes an intricate international division of labor that supplies specialized materials and products, an exceedingly hierarchical educational system for highly-expert training in physics and the material sciences, and a monopoly of resources dedicated to space travel.

Finally, it is particularly ironic that Smith’s Smart Machines and Service Work was published without an index, especially when much of this sort of work—these days—is done by machine.


Gary Roth

Gary Roth is the author of The Educated Underclass: Students and the Promise of Social Mobility (Pluto Press, 2019), and Marxism in a Lost Century: A Biography of Paul Mattick (Brill/Haymarket, 2015).


The Brooklyn Rail

JUNE 2021

All Issues