Robert Reich's foundational framework divided the American workforce into three boxes based on the nature of their work and their position in the global economy. Routine production workers performed repetitive tasks that could be codified and eventually automated. In-person service workers provided human contact that required physical presence. Symbolic analysts manipulated abstractions—words, numbers, images, code—and were predicted to capture the largest share of economic gains. The taxonomy was descriptive and predictive, identifying not merely occupational categories but power classifications. For three decades, the prediction held: symbolic analysts prospered, routine production workers declined, and in-person service workers survived at the margins. The framework became the lens through which policymakers, educators, and workers themselves understood the knowledge economy.
The three categories emerged from Reich's observation of how globalization and technological change were reshaping labor markets in the 1980s. Manufacturing jobs were disappearing, not because America was producing less, but because production was being automated or moved offshore. Service jobs were expanding, but at lower wages. And a new class of workers—the symbolic analysts—was capturing an increasing share of national income by performing work that required education, abstraction, and the manipulation of information. The classification was not arbitrary. It reflected structural differences in how each type of work related to technology, to geography, and to the mechanisms of economic competition.
The predictive power of the framework derived from its identification of scarcity as the mechanism determining economic value. Routine production work could be performed anywhere, by anyone with minimal training, which meant its value would be driven down by global competition. In-person service work required proximity but not advanced training, which limited its wage growth. Symbolic analysis required rare skills that took years to develop and could not be easily replicated—until AI learned to replicate them. The skills that commanded the highest premium for three decades turned out to be precisely the skills that large language models could perform most competently.
By 2025, the taxonomy required updating. Reich himself revised it in a PBS appearance: the categories were no longer production, service, and analysis but making, thinking, and caring. The revision acknowledged that the thinking jobs—the symbolic analysts—were now the most exposed to automation. The irony was structural: the workers who did everything right, who invested in the skills the economy rewarded, were now facing displacement by the very technology their success had funded. The three categories had served as a map. The map was being redrawn while people were still navigating by it.
The framework first appeared in The Work of Nations: Preparing Ourselves for 21st-Century Capitalism, published in 1991 while Reich was preparing to join the Clinton administration as Secretary of Labor. The book was both analysis and manifesto—an examination of how the global economy was restructuring work and a policy agenda for responding to that restructuring. The three categories provided the analytical foundation for Reich's argument that education and training, rather than trade protection, were the appropriate responses to globalization.
Routine production workers. Performed repetitive tasks following explicit rules—assembly-line work, data processing, standardized manufacturing—the category most vulnerable to automation and offshoring.
In-person service workers. Provided human contact requiring physical presence—janitors, retail clerks, home health aides, food service—resistant to automation but limited in wage growth by low barriers to entry.
Symbolic analysts. Manipulated abstractions to solve problems, identify patterns, and create value through symbolic work—the category that would capture the largest gains in the knowledge economy and the category most exposed in the AI economy.
Power classifications, not merely occupational. The three categories described not just what people did but their bargaining position in the economy—their ability to command premium wages through scarcity of skills.
The revision to making, thinking, caring. Reich's 2025 update acknowledged that symbolic manipulation—the thinking work—was now the category most at risk from AI automation, inverting the original hierarchy.
Critics argued that the three-category framework oversimplified the workforce and ignored significant variations within each category. The distinction between routine and non-routine cognitive work, which Reich did not develop in 1991, became central to understanding AI's differential impact on symbolic analysts. The framework's American focus limited its application to nations with different economic structures and social contracts.