Applied case studies
Read the evidence before the conclusion.
Each case starts with a fictional situation, separates observations from assumptions, and ends with a conclusion that preserves uncertainty. The goal is to practice reading technical claims, not to predict markets or recommend products.
Case library
Three systems, three different evidence paths
Start with the domain you encounter most often. Every case includes a worked evidence table, a repeatable checklist, vocabulary links, and first-party or standards-based sources.
Reading a Stablecoin Depeg Without Jumping to Conclusions
A hypothetical case study for separating secondary-market price, redemption access, reserve evidence, and liquidity during a stablecoin depeg.
9 min read · Updated 2026-07-29Reading an On-Chain Lending Health Factor
A worked case study for interpreting collateral, debt, liquidation thresholds, oracle inputs, and liquidity around a lending health factor.
10 min read · Updated 2026-07-29Tracing Prompt Injection in an AI Research Assistant
A defensive case study for separating trusted instructions, retrieved content, tool permissions, sensitive data, and audit evidence in an AI workflow.
11 min read · Updated 2026-07-29How to use these pages
Read once for the story, then again for the boundary.
On the second pass, cover the conclusion and write your own. Check whether every sentence is supported by the evidence table. Then compare your wording with the limits listed in “what this cannot show.” This is the same distinction-first method used in the daily puzzle and learning guides.
Read the editorial methodology