Trucking, Surveillance, and Applied Mathematics

Data Driven: Truckers, Technology, and the New Workplace Surveillance. By Karen Levy. Princeton University Press, Princeton, NJ, December 2022. 240 pages, $33.00.

Long-haul trucking is a demanding, dangerous, and poorly compensated job. When adjusted for inflation, the average yearly income for truck drivers declined from roughly $110,000 in 1980 to $47,000 in 2020. In 2019, trucking had the sixth highest fatality rate among U.S. occupations and the highest rate of serious (but not fatal) injuries. The turnover of truck drivers at large firms is close to 100 percent each year.

Truck driving is explicitly not covered by the U.S. Fair Labor Standards Act, which requires overtime pay and other forms of worker protection against exploitation. And unlike most blue-collar jobs, truckers are generally not paid by the hour; instead, they are paid by the miles that they drive. This type of payment structure creates perverse incentives, in that workers are tempted to prioritize speed over safety and spend more hours driving and fewer hours sleeping than is otherwise advisable. Yet despite its image as a haven for free-spirited scofflaws on the open road, trucking is in fact very tightly regulated.

For example, truckers are prohibited from driving more than 10 hours per day. Starting in the 1930s, drivers were required to keep written logbooks of their activities that were subject to regular inspection. These logbooks invited many forms of falsification, and cheating was rampant. In the late 1980s, large trucking companies began to utilize electronic logging devices (ELDs): small instruments that are hardwired to a truck’s engine and record a log of its engine status, location, and mileage. In 2012, Congress passed a law that mandated the use of ELDs; the law took effect in 2017. In her recent book, Data Driven: Truckers, Technology, and the New Workplace Surveillance, Cornell University sociologist Karen Levy examines the effect and implications of these devices on members of the trucking industry, overall job surveillance, and society at large.

The Federal Motor Carrier Safety Administration’s original requirements for ELDs were quite vague and provided only minimal specification of the data that was eligible for collection. As such, the many different models of available ELDs addressed the distinct objectives of various markets. On one end of the spectrum were ELDs that continually reported as much data as possible—i.e., a truck’s status, the process of driving, and even biometric information—to a corporate control center that could immediately communicate its demands to the driver. Large trucking companies generally preferred these models. At the other end of the spectrum were ELDs that appealed to the drivers themselves and only incorporated the minimum number of legally required features; though not explicitly stated, these ELDs often allowed for manipulation by the drivers.

Who benefited from ELDs? For one, ELD manufacturers now had a captive market. Large trucking companies also typically favored ELDs because they offered tighter control of their operations; in fact, these companies had largely deployed ELDs in their trucks long before the government mandate. Many smaller trucking companies, however—which constitute a large fraction of all trucking activity—generally disliked the requirement; ELDs were expensive, complicated the interactions between companies and truckers, and did not supply particularly useful information (indeed, many small companies suspected that larger corporations lobbied for the ELD mandate in order to outcompete them).

Truckers generally hated ELDs and viewed them as intrusive. Unsurprisingly, drivers soon found all kinds of ways to circumvent the devices — from crudely smashing them with hammers to disabling them or avoiding their surveillance. The ELD-based instructions from control centers were usually annoying and often oblivious to the actual situation at hand. Nevertheless, Levy remains reasonably cautious in making specific claims as to the extent of overall harm that ELDs ultimately caused truckers.

One might suppose that ELDs would appeal to government inspectors, who no longer had to worry about logbook falsification. But in actuality, their response was not clear-cut. Because the inspectors were initially unfamiliar with the many different ELD models, they had to rely on truckers to explain the devices and often simply took the truckers’ word on the ELD reports. In many cases, inspectors and truckers found themselves in a sort of alliance against the electronic devices.

The benefits of the ELD directive for the public are also unclear. The mandate sought to make trucking safer by preventing truckers from driving when fatigued, but there is reason to think that trucking accidents actually increased after it was imposed. Any safety gains from shorter driving stints were immediately lost when truckers drove faster and more recklessly to finish their routes within the allotted 10 hours.

Data Driven discusses at some length the fact that the smooth operation of modern society depends significantly on rule breaking. Levy is not at all doctrinaire on issues of regulation or technology. She does not argue that the world would be a better place if workplace regulations were abolished and the unfettered free market could operate without hindrance, or that the effects of regulation are inevitably dominated by the “law of unintended consequences;” nor does she contend that all technology or even all surveillance mechanisms merely serve to increase the power of corporations over workers. Levy maintains that the formulation and enforcement of regulations and the design and deployment of technology should be done wisely, with careful monitoring of actual consequences and a realistic and sympathetic view of the complex, multifarious realities of the situation and people involved. She also advocates for specific reforms in the trucking industry to improve the working conditions of truckers at some cost to trucking corporation profits, such as paying truckers by the hour rather than the mile.

One chapter of Data Driven addresses the possible future impact of artificial intelligence (AI) technology, particularly self-driving vehicles. Levy’s assessment is cautiously optimistic, much like that of Stanford University economist Erik Brynjolfsson. Given the political will, Levy believes that scientists can gradually and effectively introduce AI technology to society in a manner that avoids sudden disruptions and ultimately improves the efficiency, safety, and working conditions of the trucking industry.

Levy primarily carried out her research by interviewing the various actors at play: truckers, inspectors, corporate administrators, and government regulators. Her book features an interesting discussion about the techniques that she used to encourage her interlocutors to speak candidly, as well as an introspective reflection of the way in which her own personal characteristics as a young, educated, white woman helped and hindered the conversations. Levy was also visibly pregnant for a time, which had both pros and cons; people—particularly women—become more sympathetic and forthcoming, but she could no longer conduct interviews over a beer.

So, what is the role of applied mathematics in this field of inquiry? Data Driven contains essentially no math. There are numbers, certainly, but nothing that requires more than a basic understanding of simple arithmetic. Levy does not conduct any statistical analyses or computations of levels of confidence, confidence intervals, or Bayesian posteriors.

Of course, ELD technology builds on mathematics, mathematical physics, and the theory of computation in well-known ways. The internal computational mechanism of ELDs draws upon digital computation that dates back to the work of Claude Shannon and his predecessors, and the communication mechanisms stem from the theory of analogue waves by Joseph Fourier and those that came before him. A narrow but significant pathway to differential geometry and tensor theory even exists, given that ELDs use the Global Positioning System (GPS) for location and clocks in GPS satellites must account for the gravitational effects that are posited by general relativity. Behind the scenes, statisticians in the policy and planning departments of both corporations and government agencies are undoubtedly deploying statistical packages and decision support software that squeeze all possible information out of the vast quantities of data that ELDs constantly generate in millions of U.S. trucks; they presumably use this information to inform corporate strategies and government policies.

Regardless, it seems to me that the more important question that Levy’s work poses for SIAM News readers is whether and how mathematics can help achieve society’s best goals. “As digital technologies flourish across social life, they stand poised to reorient how people relate to institutions, to each other, and to themselves,” Levy writes. “Often, the best way to think about technological change is not to focus solely on the technology, but to strengthen the social institutions and relations that surround it. Only by doing so can we ensure that digital technologies become part of a vibrant social order that protects workers and promotes human dignity.” I hope and believe that mathematical thinking could be a part of this process.