• Lesson 0001 Big-picture map The whole method on one page — four contributions and how every module maps to a claim.
• Lesson 0002 The estimand & why ψ travels Why the win-rate is the probabilistic index, and why that identity is the paper's leverage.
• Lesson 0003 Inference under crossed dependence Why naive SEs are wrong when comparisons share prompts, judges, and reused responses.
• Lesson 0004 Prompt-cancellation & standardization Why the difference-form PIM can't adjust for prompt mix — and the standardized-marginal fix.
• Lesson 0005 Pairwise → leaderboard Rank-confidence sets and transitivity diagnostics: report what the data actually resolve.
• Lesson 0006 Identification & the seams to Paper 2 Where identification holds, where it strains, and the openings that become the next paper.
• Lesson 0007 Capstone: defending the leaderboard Put it together — defend every number against a skeptical reviewer.
• Lesson 0008 The standardized-marginal argument for Jean The noncollapsibility argument, made airtight.
• Lesson 0009 From ψ to the leaderboard score, and how we linearize it The influence-function derivation behind the inference primitive.
• Reference Glossary All symbols and terms in one place.