CONCEPT
Abstracted Empiricism
Mills's term for the accumulation of data without theoretical framework — the productivity-metric literature of the AI discourse is its current form, producing mountains of findings that add up to nothing because no one has asked what they are for.
Abstracted empiricism, in Mills's diagnosis, is the accumulation of survey results, statistical analyses, and measurement data without theoretical framework capable of making the findings meaningful. His 1959 target was Paul Lazarsfeld's survey-research program at Columbia, which produced rigorous data about specific questions while failing to ask whether the questions were worth answering. The AI discourse reproduces this tendency in the productivity-metric literature — twentyfold productivity gains, lines of code per hour, time-to-deployment, adoption curves, revenue per builder — whose precision captures something real while systematically failing to ask whose interests the measurements serve, what they omit, and what structural arrangements produce the distributions they describe.
In The You On AI Field Guide
The metrics show that builders are more productive. They do not show whether the additional productivity is making the builders more capable or merely more exhausted. They do not show whether the productivity gains are flowing to the builders or being captured