Imagine you run a restaurant and your signature dish is a complex mole sauce that takes three days to prepare. That dish is why people come to you specifically. Now imagine every restaurant in town starts offering the same mole. You still make it — your task count hasn't changed — but the economic meaning of that task has collapsed. Meanwhile, if you drop the mole and replace it with a burger, your menu is the same length but your distinctiveness has evaporated. This paper is about measuring exactly that kind of structural erosion across the entire U.S. labor market. The committed claim: occupational uniqueness — how distinctive an occupation's task bundle is relative to all other occupations — is a measurable, economically meaningful dimension of occupational change that existing task-based frameworks miss entirely. Using 196,563 task statements from the Dictionary of Occupational Titles and ONET spanning 1977–2018, Vestrelli and Fronzetti Colladon build a framework that tracks individual tasks across occupations and over time, separating raw task quantity changes from changes in how exclusive those tasks are to a given occupation. The key finding is asymmetric and sharp. At the occupation level, adding tasks that are already widespread across the economy is associated with lower wage growth. Removing tasks that are specific to your occupation is also associated with lower wage growth. The intuition is clean: if your new tasks are everyone else's tasks too, you've gained work but not bargaining power. If your lost tasks were what made you distinctive, you've lost the moat. Occupations that became less unique saw weaker wage trajectories. At the economy-wide level, the pattern flips in an instructive way. Adding new tasks to the economy correlates with wage growth; removing tasks correlates with wage decline. This is consistent with a story where genuine task creation (not just redistribution) is generative, while task elimination is destructive. The mismatch between occupation-level and economy-level patterns is the paper's most analytically productive result — it demonstrates that identical-looking changes in task count can reflect economically opposite processes depending on whether you account for cross-occupational task distribution. Methodologically, this sits in the Autor-Levy-Murnane tradition of task-based labor economics, but it introduces a genuinely new measurement dimension. Prior work treats occupations as self-contained bundles — you count the tasks, classify them (routine vs. non-routine, cognitive vs. manual), and track changes. This paper adds a network-aware layer: every task is positioned not just within its occupation but across the full occupational landscape. The 'uniqueness' of an occupation is then a structural property of the economy, not just a property of the job. The validation regime is correlational, not causal. The authors are careful with language, using 'associated with' rather than 'causes,' and the regressions control for standard confounders. But there's no instrument, no natural experiment, no shock that isolates the uniqueness channel from other forces reshaping wages over the same 41-year period. The dataset itself — harmonizing DOT and ONET across four decades — is a substantial contribution, but the causal interpretation remains open. The most interesting successor question is automation. If AI and robotics preferentially eliminate occupation-specific tasks (the hard, distinctive stuff) or flood occupations with widely-shared generic tasks (data entry, scheduling, basic communication), this framework predicts exactly the wage structure hollowing we've observed. The authors don't run a forward-looking AI-exposure analysis — that's the obvious next paper, and it's likely either in progress or being saved.
