Methodology
Learning in Practice
The habits behind this project — curiosity, accountability, context, and openness to challenge — made visible in the work itself.
| Habit | What it requires in practice | Where it's demonstrated here |
|---|---|---|
| Intellectual curiosity that goes looking for the counterargument | Not settling for a comfortable conclusion — actively seeking out the case against your own position | The mandatory Bear Case column in every position's Risk/Reward table and the Learning in Practice page. |
| Accountability that survives a bad month as well as a good one | Publishing results and mistakes on the same schedule, regardless of how either one looks | The Portfolio Review & Retrospective format in every full letter, and the standing Compliance Ledger that keeps a rule breach listed until it's actually resolved |
| A human-centered read on what a company's numbers actually mean | Treating a balance sheet as a description of a real business with customers and employees, not an abstract spreadsheet | Every position's Fundamental Case, and the Opinion section's editorial writing on the businesses and people behind the AI capex theme |
| Pressure-tested research over solitary conviction | Being wrong in front of someone who will tell you so, not just in front of yourself | Named plainly as a limit of this project: the work is independent, but stronger conclusions come from thoughtful challenge before action as well as after. |
The honest version of this page
A crosswalk table is easy to write and easy to overstate. The real claim here is narrower: these habits matter only when they are visible in the work — in the sources, the written rules, and the record of changed views.
A worked example, not just a table
Rather than only assert these habits, The Physical Limits of Compute is one underneath them: a full industry whitepaper on the 2026–2027 AI-infrastructure cycle, including a dedicated section arguing against my own thesis and a closing section on how the research actually changes what I hold. It is the same rigor from the table above, applied once, in full, on the record.