The following is an excerpt from Seeing Like a Supply Chain by Miriam Posner, now out at Yale University Press.
For decades, people who make Nike products in Cambodian factories have been fainting. It’s commonplace for workers to labor overtime in intense heat, compelled by demanding quotas to ignore their bodies’ need for breaks and water. Up to fifteen people per month become so dizzy or weak that they’re unable to stand. “You’re sweating profusely, walking between the buildings and in the buildings as well,” reported an outside inspector. “It was goddamn hot inside those plants, for sure.” Reporters have been flagging labor abuse in Cambodian factories since 1999, when Bill Clinton signed a trade agreement to boost garment exports from the country. Nike moved production out of Cambodia in 2000 but moved it back after receiving promises of better human-rights enforcement—though reports still regularly emerge of dangerous conditions at Cambodian garment factories.
Who gets the blame for a problem like the Nike faintings? Nike, after all, doesn’t own the factories and says it relies on outside auditors to oversee the plants’ conditions. Certainly, the factories’ managers bear some responsibility. But then again, they likely wouldn’t have made the same decisions were they not under the company’s intense pressure to produce garments on time and for a minimal cost. And yet every time this problem resurfaces, Nike’s executives profess ignorance, each time insisting that they had no idea that conditions at Cambodian factories were so abysmal.
This is far from the first time we’ve heard these claims of innocence-via-ignorance from big corporations. Like a Greek chorus, company spokespeople emerge following scandal after scandal to profess shock and ignorance. In Malaysia, Samsung and Panasonic had no idea Nepalese migrants were forced to labor under outrageous conditions. Nestlé was appalled when researchers found enslaved children within its cocoa supply chain. Apple had no inkling that workers were handling noxious chemicals with no safety equipment. Hugo Boss was shocked to discover that supervisors were holding workers captive in Indian mills. Once you start noticing, it’s impossible not to see it: time and again, a horrifying abuse of labor or the environment is “uncovered,” and time and again, a wealthy firm is aghast.
Are these companies lying? Anything is possible, but there’s a simpler, and more likely, explanation for companies’ protestations of ignorance: they’re telling the truth. It’s entirely possible to develop a lightning-quick, astoundingly efficient supply chain without having any notion of where goods are produced. In fact, this ignorance is actually a precondition for the low-cost, high-speed delivery times to which consumers have grown accustomed. Unlikely as it seems, most companies have no idea how their products get made. As a speaker intoned at an industry conference, “The truth about supply chains is that no one ever knows where the hell anything is.”
Our colossal, roiling global supply chains have evolved specifically to deliver goods with amazing speed while also maintaining a shroud of vagueness over the conditions of their production. This book shows how and why this is technically accomplished. The strategic disavowal of information, as we’ll see, extends from supply-chain professionals, down through supply-chain software itself, and to the way the software handles data. No one sets out to make Cambodian workers faint. But the arrangement of global trade makes a scandal like this inevitable—and yet, somehow, unpunishable.
Even if you don’t pay much attention to global trade, you might have encountered this pattern in a different context. In 2025, when a teenager used ChatGPT to plan his suicide, OpenAI blamed the difficulty of controlling large language models (LLMs). “We’ve learned over time that they can sometimes become less reliable in long interactions,” a spokesperson lamented. When Amazon used machine learning to screen job applicants, the tool efficiently learned to filter out resumes from women. Amazon quickly expressed dismay and distanced itself from the project. When Google Photos labeled Black people “gorillas,” the company was “appalled and genuinely sorry this happened,” and pledged to take “immediate action.” When Meta’s AI-driven content-moderation tools deleted vital evidence of human-rights violations, the company conceded that the tools may have been “overcautious.” Regretful though they were, none of these companies showed any interest in curbing the role of LLMs in their products.
And it’s true that no one thinks that anyone at OpenAI wanted the teen to commit suicide. No more did anyone at Meta specifically want to make it harder to prosecute human-rights abuses. So who do you blame when no one’s to blame? For AI systems, as for supply-chain management (SCM), “responsibility is not to be found ‘here’ regardless of where ‘here’ is. Context is perennially displaced to elsewhere,” write the information scholars David Gray Widder and Dawn Nafus. Could this shared tendency—toward predictable disaster and genuine ignorance—be a coincidence? History suggests not. The technology of the global supply chain echoes the technology of machine learning because the two have a surprising amount in common, from the assumptions they share about the way the world works to their tortured relationship to interior visibility. In both cases, technologists who dream of self-governing systems have accepted—even embraced—a persistent opacity at the heart of the mechanisms they’ve designed.
Supply-chain management is a specific, eccentric profession. Its history has never been foreordained. But it’s also bound up much more closely than we’ve thought with the larger history of computation. Given the critical importance of global supply chains, it’s worth understanding SCM in its own right. But there’s a second benefit, too: the history of SCM tells us a surprising amount about the priorities, trade-offs, and effects of the way we transform the material of reality into the bloodstream of computation. It helps us see that for all our current enthusiasm for data, surveillance, and prediction, modern computation also depends a great deal on strategically limiting what we can know.
This article has been excerpted from “Seeing Like a Supply Chain” by Miriam Posner, new from Yale University Press. Copyright © 2026 by Miriam Posner. Published by permission. All rights reserved.
Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.
Subscribe here for ProMarket’s weekly newsletter, Special Interest, to stay up to date on ProMarket’s coverage of the political economy and other content from the Stigler Center.
