Public case index

Public index only: externally verifiable cases and chapter mapping. The full case library — including internal I-series cards, oral material, and pending data — lives in the private research repository.

Inclusion rules #

• Target 10–15 key comparisons for the book.
• Every core claim needs winner and loser samples where possible.
• Report medians separately; never blend both sides into one average.
• Internal experiment cases may support the narrative but cannot be the only evidence.

Chapter coverage (external cases) #

| Chapter | Winner side (external) | Loser side (external) | Public status | | --- | --- | --- | --- | | 1 Execution zero | Duolingo, Chuan Shen pivot | Chegg | Balanced | | 2 Abundance paradox | RAND data, Ai Wei Wu case | Klarna, MD Anderson | Balanced | | 4 Judgment | Duan Yongping lineage | Wang Laboratories | Balanced | | 5 Limiting resources | Amazon capital allocation | Evergrande | Balanced | | 6 Standards | Bezos invariants, local standards | Qudian | Balanced | | 8 Interference | Xiaomi, Zhang Xue motorcycle | Chicecream (Zhong Xue Gao) | Balanced | | 9 Consensus machine | Chuan Shen, Context not Control | Nokia | Balanced | | 12 Product path | "Later but better", BBK/OPPO | Meta | Balanced |

Two-sided sample discipline #

• Collect winners and losers with equal priority.
• Freeze the win/loss definition before selecting samples.
• Match industry, stage, constraints, and time window where possible.
• Downgrade to "hypothesis" when a pair is incomplete.

How to contribute #

1. Open a GitHub Issue with sources and chapter mapping.
2. Public pull requests may only include redacted, citable material.
3. Operating data, customer names, and unaudited metrics must not appear in public PRs.

Full cards, comparison groups, and verification logs are not published here.