Every line of work on this site eventually arrives at the same question: does the instrument measure the construct it claims? This is where those arrivals meet — and where the measurement work lives as a program of its own.
In the dissertation, the instruments are information-theoretic segregation indices — built so that what readers chose is never confused with what they were served, and so that a null result is a finding rather than a shrug. In Stance Is Not a Construct, the instrument under test is the language model itself. Political Geometry asks the interior version of the same question. And before any of that comes the step the pipeline usually skips:
- Conceptualization — the step before instruments: what a concept is, and what its community means by it. First paper: What Do Experts Mean by Polycrisis? In Prep
The habit underneath all of it: test the instrument before trusting the number. It is also what I teach — methods as working instruments with known failure modes.