The Brooklyn Rail

JUL-AUG 2022

All Issues
JUL-AUG 2022 Issue
Field Notes

What Do (Digital) Bosses Do?

Managing Decline

Not long ago, the New York Times published an insightful piece of labor reporting shining a light on working conditions in the US trucking industry. It told the story of a young worker who, still living with his mother, made the decision to leave the grind and uncertainty of working part-time for ride-sharing platforms like Uber to get a commercial driver’s license. Taking a job as a long-haul trucker was appealing, for reasons one might guess: better pay and working conditions, on the one hand, the autonomy and allure of the “open road,” on the other. Of course, this young worker found neither in his new line of work. The pay was bad, the hours long and, open road notwithstanding, his truck cab came equipped with “an unbelievable level of automation and surveillance.”1 The typical truck cab is, it turns out, riddled with cameras and sensors training a close eye on the driver’s every move: these devices even track the driver’s eye movement, presumably to detect any lapses of attention. Driving a semi is, to be sure, a dangerous occupation. Long, monotonous hours mean tired drivers; a moment of distraction can mean death, for the driver or for those in his or her path. Trucking companies justify their fully-wired cabs in the name of safety. But such oversight necessarily doubles as a disciplining device, a way of scrutinizing the activity of a poorly-paid employee from hundreds or even thousands of miles distance. Putatively alone on the job, such employees can be “seen” by a boss they may never have met, but who can track and even correct—by remotely disabling the vehicle, for example—their underlings’ behavior. If drivers get tired, this boss never sleeps. In the truck cab, your boss is your co-pilot: not a god, but hardly human.

The New York Times’ anecdotal account is useful for my purposes here, because the story it tells about the use of automation in the contemporary workplace is vividly at odds with the rhetoric around automation that has taken shape over the past ten years or so, in business schools, in Silicon Valley boardrooms, and in generally credulous reporting in the mainstream press. Indeed, the story turns the breathtaking claims surrounding automation in these places completely on their head. The narrative first sketched out in the early 2010s was that in the next quarter-century or so advances in automation and machine-learning would unfold at such speed that a significant portion of the current labor force—as much as 47 percent, a 2013 Oxford Martin School determined2—stood in danger of being replaced by machines. In widely-discussed books with titles such as Rise of the Robots and The Second Machine Age, we were told that it was not just the industrial core of the economy that was susceptible to significant advances in automation. It was the much-broader and highly-populated service sector, accounting for 4 in 5 jobs in countries like the US and the UK, that was to be decimated by the leaps being made in machine -learning: not just cashiers and bank tellers, but teachers, nurses and, indeed, truck drivers might all find their labor redundant in the coming decades. This revolution in production would, of course, trigger an explosion of productivity, the likes of which had not been seen since the Industrial Revolution. But, this story lamented, it would also result in “mass unemployment,” with up to one-half of all workers replaced by robots and thereby denied access to both work and the wage.3

The picture of the trucking industry presented in the Times also stands on its head much of what I would call the “classical” narrative around automation, dating from the first decades of the factory system until the mid-1960s or so. That story told us that the mechanization and automation of production would create conditions in the workplace in which those workers not replaced outright by machines would no longer be directly involved in the material processes of production; they would, we were told, “step to the side of the production process instead of being its chief actor.”4 The nature and substance of work itself would be transformed: workers would now be needed primarily to oversee production carried out by machines. They would transcend their own condition as workers; they would become managers of the labor process. Automation meant, in one simplistic variant of this story common on the post-war European left in the 1950s, the undermining and even overcoming of the class system itself by dint of technological development alone. Yet, in the article on the contemporary US trucking industry I alluded to earlier, automation is described as being put to far different uses:

While we often think of automation and A.I. as developments that will eventually replace workers (think of Tesla’s partly automated tractor-trailer), those tools are already in heavy use in the workplace. And they haven’t replaced workers; they’ve simply been brought in to manage declining working conditions.5

Here, then, is the ironic reversal of the canonical automation narrative: rather than replacing workers, or transforming them into overseers and managers of the production process, automation is today being deployed to replace supervisors themselves, so as to better—as the article puts it so well—“manage declining working conditions.” This, it turns out, is what digital bosses do.

Over the past thirty years or so, we have witnessed a veritable revolution in the means of monitoring, tracing, and at times choreographing the bodily movements of workers. “Here,” I argued in my recent book on automation, Smart Machines and Service Work, “we have the worst of both worlds: low-tech manual labor presided over by all-seeing, ‘tracking’ eyes and ears.” I pointed to the fact that Amazon, for example, had recently patented two wearable devices to enhance the “productivity” of its warehouse employees: these devices “would use ultrasonic pulses—pitches too high for human ears to detect—to connect with inventory modules on bins to track a worker’s hands. Vibrations would communicate information to the wearer, such as alerting someone when they put something in the wrong bin.”6 I underlined, as well, that this ability to monitor and even shape behavior is hardly confined to the workplace. Activity-tracking technologies are pervasive throughout the capitalist democracies of Europe and North America, and equally so in China. Internet searches, cell phone calls, credit card purchases, the ubiquity of closed-circuit televisions, and the growing use of facial-recognition technologies now make up just part of the web of sensors, cameras and tracking devices that blanket our so-called smart cities. Employers, for their part, have long used novel forms of observation, often technologically enhanced, to compel laborers to work faster and more efficiently, to generate more output in a given period of time. But recent developments, as the example of the trucking industry suggest, allow for the penetration of these apparatuses into fields once prized for the relative autonomy they afforded workers. More specifically, these technologies allow employers either to observe and dictate behavior in situations where laborers are no longer concentrated in factory-like environments or, when they are confined to dense, controlled workplaces, to observe and measure their activity without direct human oversight. Above all, these devices are primarily being used to manage laborers who earn meager wages, and who are not classified as skilled laborers. Indeed, they are being used precisely in those workplaces in which the kinds of labor being observed cannot itself be easily automated.

The kind of tracking and control over labor I am describing is primarily being used in what is commonly if misleadingly referred to as the service sector. The service sector includes all those laboring activities that do not create a material object as an end product. These activities typically involve direct person-to-person activities, on the one hand, or carry out commercial functions that are necessary for, but not directly involved in, commodity production, on the other. In the first category, haircutters, schoolteachers, waiters; in the second, accountants, cashiers, real estate agents and security guards. This kind of labor accounts for as much as eighty percent of the labor market in high-income economies like the US and the UK. And yet, as is clear even from the cursory sketch I’ve provided here, these occupations share very little beyond a negative definition. Whether it is a question of the vast differences in skills, income levels, or these activities’ relation to the production of value and the accumulation of capital, the so-called service sector represents a field so fragmented as to be incoherent, making the concept largely useless as an analytical category.

My argument starts from the observation that much of the low-wage and relatively labor-intensive activity taking place in the so-called service sector requires significantly more supervision than is the case in highly-mechanized, capital-intensive industries like automobile manufacturing. This is a tricky argument to make, however, since highly-automated industries like automobile production typically employ large numbers of engineers, scientists, technicians, and designers, relative to workers directly involved in production. Here we are close to the scenario conjured up by the classical automation narrative: such highly-skilled employees make up a significant portion of production workers in core industries of advanced capitalist economies. These industries, by virtue of the sheer complexity and intricacy of their production processes, which often extend across multiple work sites that are sometimes separated by thousands of miles not to mention international borders and vast oceans, require legions of personnel involved in the planning and coordination of the different segments of the manufacturing process. These activities are an unavoidable feature of complex production methods; they are required by the technical and material challenges posed by producing artifacts like the automobile. As specialized and as highly-remunerated as they might be relative to those workers directly involved in material fabrication, engineers and planners might be considered productive workers insofar as they carry out functions necessary for the production of what Marx would call “use-value.” They should not be confused with what I above called “supervisors,” who play no direct role in the creation of new use-values. This is an essential distinction whose implications I want to examine at some length in this essay.

Supervisors do not participate in the production process, even indirectly. They are, instead, tasked with monitoring and disciplining workers. They manage the labor force, not the labor process. Their mandate is to compel those workers they oversee to perform their labor at a pace and intensity sufficient to produce more “value”—note, I am no longer speaking for the moment of “use-value”—in a day’s work than the value they consume in wage goods (food, clothing, housing, health care, and so on) each day. They compel them to do so by imposing discipline on them: they monitor their performance, they hold out the promise of promotion, they threaten them with termination, all in the interest of coercing as much output as possible from a given worker during a given period of time. Such supervisors occupy a double, or divided, position within the class dynamics of the workplace. On the one hand, they are workers who are paid wages and represent a part of an employer’s labor costs. On the other hand, they represent and enforce the interests of business owners, when those business owners do not directly oversee the labor they hire. Supervisors perform the labor of exploitation.

It is perfectly reasonable to assume that, in certain situations, the same person might perform both of these functions: planner and supervisor. It is often the case that the planning and design of a labor process will incorporate disciplinary or “supervisory” considerations. In the actual practice of private producers, these operations can be hard to tell apart. I will return to this point in a moment. Analytical clarity, however, requires that we separate them as strictly as we can. However formidable the prospect of separating these functions out in a given empirical scenario, planning and supervision relate to the production of value in distinct, indeed, opposed ways. The labor of planning exercises, to be sure, a certain authority: Marx does not shy away from describing the planner’s as a “governing will” whose objective is “the interconnection and unity of the [production] process.”7 There is no coercion, however, implied by this function of “command.” Such oversight is a technical requirement of the labor process itself. Though managers of this sort play no direct role in the handling of materials or the fabrication of an object (or the performance of a service), the labor they perform should be considered an essential contribution to the collective labor employed in the production of a given use-value. The labor of supervision, however, is employed not in view of the production of a use-value, but in order to compel the creation of a maximum amount of surplus value from a given quantity of labor-power. It is not a technical requirement of a complex production process. The labor of supervision is an example of what Marx calls unproductive labor.

Productive and Unproductive, Once Again

Though the distinction between productive and unproductive labor is a central concern of the classical political economy Marx’s scientific writing criticizes so completely, its role in Marx’s value theory is often perceived as both marginal and obscure. Its most thorough treatment is found in the notebooks we know as Theories of Surplus Value, and in a chapter from the manuscripts Engels published as Volume 2 of Capital. What is more, Marx never fully assimilated his reflections on this conceptual polarity into the broader movements of his thought. The fragmentary state of his surviving scientific writing requires that we begin this labor ourselves. For the moment I will emphasize two important theoretical questions: First, one of the developmental tendencies of the capitalist mode of production not fully spelled out in Marx’s mature thought is what we can call, following Paul Mattick and Fred Moseley, the rising ratio of unproductive to productive labor employed. Just why more and more labor in advanced capitalist economies must be allocated to unproductive relative to laboring activities is a question I will address in a moment. Since unproductive labor does not itself produce value, it represents a cost to capitalists, who must pay for it out of the surplus value created in the production process. This cost is therefore paid out of the profits of business owners; it represents a drag on capital accumulation. Second, the distinction between productive and unproductive laboring activities is essential to Marx’s thought for reasons that are not only theoretical but political as well. Productive activities like those performed by engineers and managers are required for any collective labor process under particular technological and material conditions, that is, at a given scale and complexity of production. On the contrary, unproductive activities are those functions necessary solely for the accumulation of capital and, as result, represent the differentia specifica of the capitalist mode of production. Unproductive activities are the material embodiment of a particular social form of production: the production of commodities by means of the employment of wage labor. The organization of social production in non-capitalist societies would require the application of scientific knowledge in the formatting of production, as well as the carrying out of management functions—note, I do not say “managers”—in the planning of production. It would not, however, require cashiers, bankers, insurance agents, human resources departments, legal representation, or security guards. Or bosses, digital or “in person.”

To understand what Marx means by the productivity of labor, we must first register the fact that “productivity,” like many of the key categories he uses to construct his conceptual architecture, is originally split into two: it has, like the concepts of “commodity” and “labor,” a dual character or nature. Productivity, for Marx, can be viewed in concrete or technico-material terms, as the production of physical units (or “use-values”), and it can be conceived exclusively as the production of value. It might appear that Marx is concerned in Capital with the second understanding of productivity, insofar as he is preoccupied in this book with offering a theoretical account of the specifically capitalist organization of social production. This is not quite right. If Marx begins from the originally divided nature of his theoretical categories, it is only to seek their antagonistic unity. The commodity is necessarily conceived of as the contradictory coincidence of use-value and exchange-value. In the same way, concrete and abstract labor are not two types of labor, but the same labor considered from two opposed perspectives, that of the production of use-value and that of value. For Marx, only “abstract” labor—labor considered apart from its particular kind of product—is labor productive of value. The implications of this proposition are vast. We will only touch on a few of them here.

When Marx speaks of productivity in concrete terms, then, he means the production of what we can with caution call use-values. Though Marx’s terminology slips here and there in his presentation of the notion of use-value, he often makes a clear distinction between an “object of utility” generally speaking—some useful product of labor—and use-value properly speaking: a “use-value” is not a transhistorical category that recurs across different social forms of production, but is strictly speaking a determination of the commodity-form in its dual nature, i.e. the antagonistic unity of use- and exchange-value.8 So when speaking of the production of use-values, we should be sure to underline that we are still speaking of commodity production, only examined through the lens of the technical and material conditions of production. We are concerned, then, with the productivity of labor in concrete terms, i.e. how much of a particular use-value that can be produced with a given quantity of labor, only insofar as this determination of productivity enters into contradiction with its other determination, the production of value.

It is useful to contrast this way of conceiving and measuring labor productivity with the prevailing methods in mainstream economics and business journalism. Productivity is a relative term: one production unit is more or less productive than another. We measure rates of productivity: how much of something can be produced with a given unit of labor. The most successful companies typically produce more output with a given quantity of labor. But this how much is itself a problem. If we measure this output in particular physical units, like pairs of shoes, there are no issues. If we try to compare the labor productivity of different lines of production, however, we get into trouble: how to compare the productivity of a childcare worker with someone who works in a shoe factory? In a sense, the only way to compare their output is to measure each in money terms. Only such a value-measure of productivity can impose a unity on the disparate concrete use-values of the myriad lines of production typical of highly ramified social division of labor. Of course, this process of comparing and measuring different productivity rates of otherwise incomparable concrete labor processes happens all of the time: this is precisely what happens when goods are priced for sale as equal in value.

This method of measuring productivity, however, is beset by its own conceptual and theoretical problems. For my purposes, I will focus on one in particular: a large fraction of the labor activity taking place in the advanced capitalist economies “produce” an output measurable in money terms but produce no use-values at all, in Marx’s sense. This is particularly true of what we can call “circulation labor,” i.e. all those activities—the work of buying and selling—that are necessary to realize (as Marx puts it) the value created in the production process but which as a result produce neither use-value nor value themselves.9 These functions require both time and labor, and must be paid out of the eventual profits of producers (or, more strictly put, out of the surplus value generated through the exploitation of labor-power in production). Circulation labor, like the work of a cashier, does not produce new value. It transforms one form of an existing sum of value into another: from the commodity- to the money-form (C-M) or vice versa (M-C). Such activities are “form-determined,” to use Marx’s technical term: they reflect the particular conditions of the social form of the product of labor, the commodity, whose distribution is mediated by market exchange.

Productive labor in Marx’s sense must contribute to the production of value and use-values both. I emphasize this dual nature of productive labor because, in addition to circulation labor, there is an even larger fraction of the laboring activity in advanced capitalist economies that is performed in the so-called public sector. Consider this: over forty percent of the United Kingdom’s GDP is made up of government spending. Such spending is composed of wages paid to state employees (police officers, legislative aides, postal employees, etc.) and of commodities produced by the private economy that are directly purchased by the state.10 Consider teachers. Teaching clearly produces a use-value, and is generally considered a “service” because it does not directly transform some raw material in the way a dress-maker does. The category of “services” is next to useless for our considerations, however, because the term describes the nature of the product, rather than the part played by the labor employed in value production. Many services, but by no means all, produce use-values. Teachers and masseurs do; cashiers, insurance agents, and security guards do not. Public school teachers are paid wages, and therefore are laborers just like those who produce shoes in a factory. But they do not produce commodities. For the most part, the services teachers provide are not sold directly on the market. They, therefore, do not produce value, nor are their activities regulated by the law of value in the way that private, competing producers’ activities are. Here, then, we have a clear example of a very different type of unproductive labor than that we identified with circulation labor. The former produces use-values, but no value; the latter produces neither. Yet each represents a cost to capital, that is, to the private economy as a whole, insofar as these two types of labor are paid out of the total pool of surplus value produced through the exploitation of labor-power in the private economy.

Put simply, these expenditures represent a significant drag on capital accumulation. More specifically, these two “spheres” of the economy exert considerable downward pressure on the profit rate, precisely because they must be conceived as deductions from the surplus value private companies create in production. Each of these spheres—state expenditures and circulation cost—expanded dramatically in the twentieth century, long after Marx tried to theorize their relation to the accumulation process. Since the 1930s, state expenditures have not only risen precipitously, they have been conceived by states as a mechanism to offset the severest effects of accumulation crises, by “artificially” stimulating aggregate demand in periods of pronounced recession or depression. Yet these expenditures, which remained elevated even during the post-war boom, have had the long-term effect of slowing down accumulation, creating the conditions for the very crises they are meant to hold off or offset.11 The explosion in circulation labor, on the other hand, is an unavoidable consequence of the rapid rise in labor productivity in the industrial core, on the one hand, and of the extension of markets, on the other. As more and more output is produced by the private economy specifically to be sold on the market, more and more market transactions occur. Insofar as the “productivity” of this circulation labor lags behind that directly employed in production, an increasingly large share of the total labor force will be required to perform labor which represents an outright cost to capital.

What Bosses Do

I have taken great pains here to lay out Marx’s distinction between productive and unproductive labor. In doing so, I have singled out one type of unproductive labor, the labor required to perform commercial activities like buying and selling, which he calls “circulation labor.” Marx characterized such actions as the conversion of one form of value into another (money to commodity, or vice versa), rather than the creation of new value. I noted, as well, that the distinction between productive and unproductive is important not only because unproductive labor represents a drag on accumulation—this was examined above—but because unproductive functions represent activities required by the social production relations specific to the capitalist mode of production. Only when production is organized in such a way that private producers compete with one another, and the social product is distributed by sale in the market, are commercial activities needed to realize the value incorporated in commodities. In short, only in societies in which the social product takes the form of commodities is the labor of circulation necessary.

But Marx is equally concerned with another type of labor that reflects a particular form of social production: the supervisory labor we started from. In the third volume of Capital, Marx defines this labor in the following terms:

[The] work of supervision arises in all modes of production that are based on the opposition between the worker as direct producer and the proprietor of the means of production….and is [therefore] indispensable in the capitalist mode of production, since here…the production process is at the same time a process of consumption of labor-power….12

When Marx speaks here of the consumption of labor-power, he underlines the fact that for the capitalist, labor is a commodity, purchased on the labor market; like all commodities, it has an exchange value, expressed in the wage, and a use-value, the peculiar fact that its consumption creates value. The objective of supervisory labor is to ensure that the value generated by means of the consumption of labor-power exceeds the value represented by the wage, that is, the cost of reproducing that labor-power. It is the wage-form itself, the fact that labor is allocated by means of the market and assumes the form of a commodity, that requires this function of oversight. Without the constant pressure to maximize productivity imposed by the capitalist’s agent on the shop floor, the supervisor, there is nothing that compels the labor-power purchased on the market to produce any value at all, much less a sum of value over and beyond the costs of reproduction. Here, the specific form of capitalist production relations—labor assuming the form of labor-power, purchased with the wage—requires capitalists to purchase additional labor to impose discipline on this labor sufficient to produce surplus labor.

In the course of sketching out his theory of supervisory labor, Marx adds an important qualification. Concerning the “opposition between the owner of the means of production and owner of mere labor-power” prevailing in the capitalist mode of production, Marx writes: “the greater this opposition, the greater the role that this work of supervision plays.”13 What can this possibly mean? How can the opposition between capital and labor be “greater” in some circumstances, lesser in others? Opposition, antithesis, and even antagonism between capital and labor is a structural feature of the capitalist mode of production, in which those compelled to sell their labor-power own, precisely, nothing (other than their capacity to work). But we can speculate. When Marx imagined the future development of capitalism, he envisioned an ever-increasing integration of workers into massive, machine-driven productive complexes in which laborers would be disciplined, as it were, by these very machines. The “technical subordination” of the worker to the machine, Marx argues, would “give rise to a barrack-like discipline…bring[ing] the…labor of supervision to its fullest development.”14 By “fullest development,” Marx means two things that seem, on the surface of things, opposed. On the one hand, in the automatic factory—in which the organization of the labor process requires the submission of the worker’s body to the machine’s movements—we encounter a clear separation between laborers and supervisors. In the earlier period of manufacture, in which many aspects of the labor process were still controlled by highly-skilled craftsmen, the distinction was less pronounced: those who labored owned their own tools and could to some extent work at their own pace. Often their work was not directly overseen by a “supervisor” at all. This changed dramatically with the factory system. But the labor of supervision is fully developed in the automatic factory in a second sense as well: the technical subordination of the worker to the machine means that the machine itself dictates the pace and intensity of labor, diminishing the need for supervisory labor. The technical requirements of production tend to coincide with the social requirements of control. In this sense, the opposition Marx suggests can be more or less antagonistic is expressed in the degree to which a worker or workforce is subordinated to a labor process. Insofar as the machinery employed in the labor process embodies, in material form, the dictates of the owners of that machinery, the need for supervisory labor is diminished.

In Smart Machines and Service Work, I argued that supervisory labor should be understood as additional labor required by business owners to exploit other workers. Such labor is “necessary” under capitalism, but it would no longer be required if there were no separation between those who possess the means of production and those who carry out the immediate tasks of production. In this sense, supervisory labor is unproductive labor: it represents a feature of the production process—the oversight and disciplining of hired labor-power—peculiar to the capitalist mode of production. To this labor I have here opposed the labor of management, which carries out the planning and integration of complex labor processes that is a requirement of all collective labor. In an especially insightful critical response to my book, Bo Harvey argues that it is “particularly puzzling” that I would characterize supervisory labor as “unproductive,” since

the production of surplus value involves coercion and discipline—via, for example, increased intensities of work within a given time frame due to discipline or surveillance—no one should be surprised when even the most banal technological advancement includes some coercive or disciplinary aspect….Innovations in workplace discipline and surveillance simply are productive innovations insofar as they create the conditions for an increased intensity of work. If one accepts the legitimacy of relative surplus value as an analytical category, disciplinary control and the productivity of surplus value are two sides of one coin.15

For the most part, there is little to object to in this formulation. We need only recall that Marx himself described the technical subordination of the worker to the machine as the “fullest development” of the labor of supervision. Earlier I noted that the role of manager and the role of supervision are often carried out by one and the same person. I also noted, more to the point, that those who plan and organize the labor process do not simply have technical and material considerations to consider. Efforts to increase labor productivity must be reconciled with the need for maximum control over the labor process. It is not, however, hard to envision scenarios in which these two imperatives would enter into conflict, and in which the owners of the means of production would prioritize workplace discipline and control over efficiencies in the production process.16

For the most part, however, they do not have to make such choices. Refinements of the technical and material features of production, resulting in rising output per laborer, tend to double as enhancements in disciplinary methods. Marx himself makes this point especially clear in his discussion of supervisory labor in volume three of Capital. “The work of supervision and management,” he writes,

insofar as it arises from the antithetical character, the domination of capital over labor…is also directly and inseparably fused [verquickt], under the capitalist system, with the productive functions that all combined labor assigns to particular individuals as their special work.17

Marx’s use of the expression “directly and inseparably fused,” however, requires an important qualification. He is not suggesting that, at a conceptual and analytical level, the labor of supervision is a “productive” function, as Harvey suggests. To the contrary, he underlines that such labor is necessitated solely by the “antithetical” character of capitalist production relations. “Under the capitalist system,” as Marx puts it, technological innovation and class domination are, for all practical intents and purposes, one and the same. If we consider the category of relative surplus value, which Harvey rightly invokes in the passage I cite, this makes perfect sense. Relative surplus value refers to a particular method of increasing the amount of surplus value produced by a given unit of labor. If absolute surplus value is produced simply by extending the working day, the production of relative surplus value means that more surplus value is produced by technologically transforming the production process, to raise labor productivity.

The fusion of these two functions, however, is a practical unity, not a theoretical or analytical one. The danger in directly identifying technological necessity with class coercion is that, in doing so, we collapse the distinction between them, identifying technological transformation of production with class domination per se. This is precisely what Marx warns us against in the Grundrisse, when he observes that though the automatic factory is a form of production “posited by capital itself and corresponding to it,” insofar it represents the most effective means for appropriating labor time, it “does not follow that therefore subsumption under the social relation of capital is the most appropriate and ultimate social relation of production for the application of machinery.”18 Here, Marx is at pains to emphasize that though the factory system is “posited” by capital—created in response to capital’s own “needs”—there is no conceptual identity between them. Marx pointed to the existence of the cooperative factories of the Owenite movement as an example of the possible socialist use of the factory form, a mode of producing in which “the antithetical character of supervisory work disappears.”19 In such a factory, the labor of planning and management would carry on. It might be performed by particular persons designated as “managers”; this technical division of labor might assume a hierarchical form. But the nature of the authority prevailing in such a workplace would be totally other than the specific form of domination at work “under the capitalist system.”

Automation, Crisis, Control

The working conditions prevailing in the advanced economies of Europe, North America, and increasingly East Asia are quite different from those Marx described and anticipated in his analysis of the automatic factory. Over the last two decades in particular, workers in these regions have been pouring into the bottom end of the labor market: into low-wage, low-productivity service work. Indeed, since 2000 or so, and especially since the crisis of 2008, the sole significant job growth in these regions is at the bottom end of this sector: in retail, in restaurant work, in childcare, or in the most menial occupations in the health care sector. This work is typically difficult to mechanize or automate, due to the nature of the labor itself, and because the labor is so poorly paid that employers have little reason to invest in pricey fixed capital to replace it. It is in just such situations, in which labor-power is employed without the mobilization of large amounts of fixed capital—in labor-intensive, low-wage service work in particular—that the opposition or antithesis between capital and labor is heightened, and so with it the need for supervisory labor. Since supervisory labor represents an unproductive cost to capitalists, the swelling of this supervisory layer in much of the economy exerts downward pressure on the rate of profit, since this additional labor must be paid directly out of the surplus value extorted from those employees being surveilled and disciplined. Of course, since the raison d’être of this labor is to “sweat” or intensify the labor it oversees, thereby raising the “productivity” of each unit of labor and the rate of exploitation, some of the cost of this labor to the employer (or to the capitalist class as a whole) can be offset by the additional surplus value created by the discipline it imposes.

The increasing automation of this supervisory function in industries like trucking and logistics represents a quantitative extension of the field of surveillance in the workplace, insofar as it allows employers to penetrate spaces—the truck cab, for instance—once largely free of direct oversight. It represents, as well, a qualitative leap in the nature of oversight, offering real-time monitoring of the bodily movements of laborers—including something as “inward” as attention itself—to a degree never before approached in the history of workplace discipline. Yet the urge to replace in-person bosses with their digital stand-ins must be seen as a response to the broader crisis of “productivity” plaguing the advanced capitalist economies. On the one hand, the expansion of low-wage, labor-intensive sectors of the economy means that employers must hire more and more supervisory staff—far more than in highly industrialized settings.20 On the other hand, significant labor productivity gains are difficult to come by in this low-wage sector, since it employs little labor-saving machinery. Confronted with this dilemma, business owners find themselves compelled to automate not the labor process itself, but the supervisory labor overseeing it. Such a “digitization” of the supervisory function both economizes on the additional outlays in wages required by heightened workplace surveillance, while also extending the opportunities for oversight in ways unthinkable half a century ago. The digital boss arises in crisis conditions peculiar to the twenty-first century, in which productivity gains are more often than achieved not through the automation of more and more workplaces, but through leaps in the extent and intensity of workplace surveillance.

In his vision of the socialist use of machinery, Marx held out hope that “the forms developed in the womb of the capitalist mode of production may be separated and liberated from their antithetical capitalist character.”21 Of course, he had the factory system in mind. The grim history of the twentieth century suggests such a separation and liberation would prove more difficult than he imagined. But in the twenty-first century, the technological marvels are different in kind and present other dilemmas. The rise of the automated or digital boss proposes no emancipatory horizon; the sensor-studded truck cab will not be liberated from its antithetical capitalist character. The factory system represented for Marx both the complete realization of the capitalist form of social production and the promise of a form of cooperative labor planned and carried out by the producers themselves. The new regime of work being implemented in the trucking industry, the Amazon warehouse, and elsewhere is far bleaker. The digital management of cheap labor, whether in the service sector or in poorly paid, labor-intensive occupations that store and move goods, represents a form of not relative but absolute surplus value extraction, whose limit is finally the capacities and frailty of the human body itself. Absolute surplus value is typically thought of as extensive in nature, involving the prolongation of the working day. But it can also be intensive, compelling the production of more value not by lengthening the working day but by speeding up the pace of work. The history of capitalism in the nineteenth century already demonstrated that whereas the use of technological innovations to raise productivity has no intrinsic limit, the intensification of labor encounters a limit in the broken bodies of the laborers.

It is perfectly reasonable to characterize the portrait of the current capitalist class dynamic I have sketched out above as dire, even demoralizing. The image of the digital boss represents forms of control over the labor process and over the workplace that are unprecedented and, in the eyes of many observers, “dystopian.” But we should also see this tendency as the expression of a deep, and unrelenting, crisis of accumulation. It represents the hidden, dark side of the automation narrative launched in the wake of the 2008 crisis, maybe even what Hegel would call its “truth.” What it expresses is a fundamental dilemma for the capitalist class, who preside over mature capitalist economies which have, since the 1970s, exhibited little to no dynamism: no growth in GDP, labor productivity, real wages, and so on. The world of work has been transformed dramatically over the past fifty years, as the centrality of the factory and its form of discipline have faded, in the West at least. The antithetical relations between capital and labor in the workplace, in a world dominated by low-wage, low-skill service labor, have intensified, rather than diminished over this period, as Marx might have imagined. The overcoming of this antithesis can no longer be imagined as the outcome of technological development itself, nor can workers imagine themselves assuming the functions of planning and management of the labor process, as the factory form born in the womb of capitalist social relations is liberated from all forms of domination. The prospects for struggle that emerge within this new tendency toward the production of absolute surplus value are obscure and uncertain. Today’s responses to the antithesis between capital and labor are negative in nature, such as the systematic sabotage of the means of surveillance and the wholesale refusal of wage labor. The alternative, however, seems to be the physical and mental degradation of the laborers themselves.

  1. Robin Kaiser-Schatzlein, “How Life as a Trucker Devolved Into a Dystopian Nightmare,” The New York Times, March 15, 2022;
  2. Carl Benedikt Frey and Michael Osborne, “The Future of Employment: How Susceptible Are Jobs to Computerisation?,” Oxford Martin School, September 2013.
  3. One of the exemplary objects of this narrative has been self-driving cars, whose advent was hailed by many of the stock markets’ tech-sector darlings as imminent. Self-driving automobiles would represent a signal innovation for an automobile industry notable for its stagnation: there is little substantive difference between a Chevrolet built in 1920 and one made last year. The promised revolution, however, would not simply be in the consumer market for cars, but in public transportation, on the one hand, and the trucking industry on the other. The ride-sharing platform Uber emphatically billed itself as a technology start-up whose ultimate ambition was the development of autonomous automobiles that would not only replace existing forms of urban transportation (taxis, buses, subways) but would remake the edifice of urban space itself. In 2020, Uber announced it would be giving up its efforts to develop autonomous vehicle technologies.
  4. Karl Marx, Grundrisse: Foundations of the Critique of Political Economy, trans. Martin Nicolaus (Harmondsworth, 1973), p. 706.
  5. Kaiser-Schatzlein, “How Life as a Trucker Devolved Into a Dystopian Nightmare.”
  6. Jason E. Smith, Smart Machines and Service Work: Automation in an Age of Stagnation (Reaktion, 2020), pp. 110.
  7. Karl Marx, Capital: A Critique of Political Economy. Volume 3, translated by David Fernbach (Penguin, 1991), p. 507.
  8. The fatal contradiction between the production of use-value and the production of value (and therefore, surplus-value) will, once fully developed in Marx’s account, assume a final form in the contradiction between the rising productivity of labor, brought about by ever-higher ratios of machinery to living labor, and the resulting decline in the rate of profit. This contradiction is the ultimate source of the crisis tendency of the capitalist mode of production
  9. Marx’s argument that value does not pre-exist the act of exchange—the so-called “monetary theory of value” proposed by Michael Heinrich and others—in no way obviates the distinction between production and circulation. It is equally true that production undertaken privately is only validated as constituting a part of social labor (“abstract labor”) in the exchange process and that circulation labor does not produce value, but only converts one form of value into another. Marx’s entire theoretical edifice depends on this distinction. What his “monetary” theory of value confirms, however, is that a. value never appears outside of its phenomenal form, “money,” while remaining conceptually distinct from it, and b. that while value is produced only in the production process, the amount (in certain cases=0) of value-producing labor actually taking place in a given workplace is only determined retroactively, when these activities are compared through and at the moment of exchange.
  10. The state sector includes the wages of all government employees: those who work for, say, circuit courts, the police, the post office, the military, legislative aides, and so on. But the state sector also includes all of the purchases the state sector makes of products made and marketed by the private sector, since these products—including, for example, subsidizing health care costs, a huge part of the US GDP—are paid for not out of capital, but out of state revenue primarily generated by taxation, and which represents a deduction from the total surplus-value generated by the private economy.
  11. See Paul Mattick, Jr., Business as Usual: The Economic Crisis and the Failure of Capitalism (Reaktion 2011), especially chapters 4 and 5.
  12. Capital, volume 3, pp. 507-8.
  13. Ibid, 509, 507. My italics.
  14. Capital, volume One, p. 549.
  15. Bo Harvey, “Harvey on Smith, ‘Smart Machines and Service Work,’” H-Socialisms (August 2021);
  16. In his notorious “What do Bosses Do?,” Stephen Marglin went so far as to say that the capitalist reorganization of production was actually less productive than prevailing production methods, and was undertaken not in the interest of technological “efficiency”—the most economical use of existing labor inputs—but in order to secure class domination in the workplace and to establish enduring patterns of capital accumulation. See “What Do Bosses Do? The origins and functions of hierarchy in capitalist production, Part I,” The Review of Radical Political Economics 6:2 (1974): 60-112.
  17. Capital, volume 3, p. 510. My italics. Note that Marx combines the categories of “supervision and management” in a way that I, for reasons of conceptual clarity, do not.
  18. Grundrisse 699-700; Marx famously speaks of a capitalist “use” of machinery in Capital vol 1, chapter 15.
  19. Capital, volume 3, p. 512.
  20. Of course, by “more” we mean more per unit of labor.
  21. Capital, volume 3, p. 511.


Jason E. Smith

Jason E. Smith writes about contemporary art, philosophy, and politics. His book, Smart Machines and Service Work: Automation in an Age of Stagnation (London: Reaktion, 2020) is part of the Field Notes series.


The Brooklyn Rail

JUL-AUG 2022

All Issues