How the Series Began
The Formula of Recurrence started from a single question: when productivity explodes, why do some people get displaced while others read the moment and take the opportunity?
From Rome's latifundia to the factory system of the Industrial Revolution, and now to the AI revolution unfolding in real time — I kept running into the same pattern, repeating across two thousand years. The six volumes are the attempt to lay that pattern out carefully enough that a reader can see it, name it, and act on it.
Writing Timeline
Tracing the pattern of productivity explosions across three eras — Rome, the Industrial Revolution, and AI — in 18 chapters. Korean and English manuscripts were developed in parallel from the start.
Primary Research ClaudeA parallel analysis of the Silicon Valley and Shenzhen AI ecosystems — semiconductor supply chains, US–China decoupling, and the strategic position of third countries caught between the two blocs. Three infographics were hand-built for this volume.
Field Research ClaudeVolume 3 traces six hundred years of capital-allocation mechanisms, from the Medici bank to DeFi. Volume 4 runs from the Roman Senate to the EU AI Act, diagnosing the structural reasons institutions cannot keep up with technology. My years as head of strategy at a Korean savings bank gave Volume 4 its operational backbone — these are problems I watched play out from the inside.
Claude CodexVolume 5 was written under the most disciplined process of the series:
Phase A — Structure. A BOOK_STRUCTURE.md file laid out the full argument arc across 19 chapters before a word of prose was written.
Phase B — First draft. Seven country-by-country "mirror" analyses: Taiwan, Israel, Singapore, the Netherlands, Indonesia, the UAE, and Japan.
Phase C — Fact check. All 435 sources independently verified; data points cross-checked for accuracy.
Phase D — Quality review. Argument structure and narrative coherence audited end to end.
Phase E — Epilogue and appendix. Including an investment framework built on the volume's findings.
Phase F — Consistency pass. Policy boxes reinforced, cross-references corrected, terminology unified across chapters.
The final volume in the series. The binary pivots from "the displaced vs. the discerning" to "the self I'm leaving behind vs. the self I'm becoming," asking harder questions about work, identity, and what a human being is for in an age of capable machines.
Claude Opus Personal ExperienceHow AI Tools Were Used
This series was not ghost-written by AI. It was written with AI, in specific and deliberately limited roles — as a research assistant, a fact-checker, and a structural auditor. Every argument, every narrative choice, and every judgment call in the text is the author's. The model was allowed to do the following and nothing else:
- Research acceleration — rapid gathering and organization of academic papers, reports, and statistical sources.
- Fact checking — cross-verification of dates, numbers, and quotations against their original sources.
- Structural audit — stress-testing the logical flow of an argument and surfacing gaps or non sequiturs.
- Translation first draft — generating the initial English draft of Volume 1 (then reworked from scratch by the author).
Data and Source Methodology
Every data point and citation used across the series was selected and verified under the same rules:
- Primary sources first. Whenever possible, original reports, statistical releases, and legal texts were consulted directly rather than through secondary summaries.
- Cross-validation. Core figures were confirmed against at least two independent sources before entering the manuscript.
- Explicit as-of dates. Every figure carries its reference date — essential in fast-moving domains like AI and semiconductors, where "current" means "current as of when."
- Honest uncertainty. Estimates, projections, and opinions are labeled as such. The reader should always know when they are being shown a measurement versus an interpretation.
The bibliography for each volume is available on its own page.