How do you tell whether someone understands something, rather than just sounds like they do? You design and run the instruments that answer that question — starting with your own claims, ending with a case no instrument can settle.
What you will do
Every week you build and run a small instrument against a piece of reasoning — yours, a classmate's, or a fabricated one — and report what it detected, what it missed, and what it couldn't tell you at all. The artefact you leave with is not a portfolio of answers; it's a record of which of your own claims to understand something actually held up under a test you designed yourself.
Who it is for
Anyone who has caught themselves saying "I get it" a beat too quickly — after a lecture, a tutorial, a fluent answer from an AI. No prior subject knowledge is assumed, because the course is domain-neutral: you supply the material you already half-know, and the course supplies the pressure. It will suit you badly if you want a settled answer by the final week; that's not on offer here.