Real decisions from real readings, with the working shown — including the mistakes.
Each piece takes a question that came up while producing an actual published reading, tests it, and reports what changed. The charts are real, the figures are recomputed, and where the honest answer was 'this does not settle it', that is what it says.