Method & data

How the guide was built.

Enough detail to understand what the numbers mean, where judgment entered, and what this project cannot claim.

Generated September 2, 2026

01 · Source

Indexed transcript set

Spoken.md purchased All-In transcript archive. The local archive contains 406 speaker-labeled episodes from 2020-03-15 through 2026-08-29. Raw transcripts remain private; this guide publishes derived analysis and brief source-linked examples.

Public transcripts
406
Words
5,813,291
Date range
Mar 2020Aug 2026
02 · Analysis

Recent first, history second

The primary editorial window is the trailing 12 months beginning September 1, 2025. Conversation ordering gives extra weight to episodes in that window while retaining the full history as context.

Primary window
132 episodes
Earlier baseline
274 episodes
03 · Glossary

Terms grounded in transcripts

A curated catalog of 126 candidate concepts was matched against full transcripts. Published terms have at least one matching episode and are ordered by distinct episode coverage, with extra weight on episodes from the latest year. Optional examples show transcript context.

04 · Limits

Read claims with care

  • Automated transcript wording and speaker attribution can contain errors.
  • Phrase counts indicate attention, not endorsement or sentiment.
  • An episode can belong to several themes.
  • Guest-heavy and Summit episodes may reflect guest views rather than recurring show themes.

Download the work

Inspect or build on the data.

Machine-readable analysis, episode-level scoring and trend summaries are available for community use.