About

1. A finder pass reads every post and logs every forward-looking statement with a verbatim quote; a script verifies each quote against the cached text.
2. Two independent coders map each statement to a registry eventA proposition written from a template on the asset, with a resolution criterion and a named public source, before the outcome is looked up. The same event resolves every person's items on it. Method › and a deadlineThe date by which the event must happen for the item to be true. It comes only from the person's words through the anchor table. Method › through the anchor tableThe fixed table that turns the person's own words ('by 2026', 'Q2', 'within two years') into a deadline. The coder never invents a date. Method › and set the probability binOne of five lexicon steps: A 'will' 0.90, B 'probably' 0.70, C 'may' 0.50, D 'unlikely' 0.30, E 'never' 0.10. A stated number goes to the nearest bin for calibration only. Method ›; a mismatch on event or deadline voids the itemAn admitted statement with its intake fields: event, deadline, probability, tags. An item carries no outcome fields; the outcome lives on the registry event. Method ›.
3. The intake file is frozen with a hash before any outcome is looked up.
4. A resolver who never sees the probabilities resolves each registry event from primary sources; a rechecker argues the opposite verdict.
5. Scores are computed at build time from the JSON data; the committed snapshot is drift-checked.

The full archive of The Frontier Psychiatrists on Substack, every post from 2021-05-06 to 2026-09-11, free and paid, read in full through the Substack API (free) and the subscriber session (paid). Restacks of other authors are excluded. A second sweep on 2026-09-14 added verbatim quotes from podcast transcripts, press and other channels found by search.

• Substack archive API · 1,126 posts: 927 free, 176 paid, 23 founding
• Off-site sweep (2026-09-14): podcast and video transcripts, press quotes and posts outside the Substack, found by search; every quote checked on its page · 71 statements added; the Substack archive stays the tier-A corpus

Ad hoc collection from public statements (own posts, verbatim interviews, dated press quotes). Rows are shown and counted; no headline number until the declared tier-B corpus (own Substack, X and LinkedIn posts, atai shareholder letters, verbatim transcripts) is read in full.

• Own Substack
• X posts
• atai shareholder letters and press releases
• Interviews and podcasts with verbatim transcripts
• Second sweep (2026-09-14): own Substack read in full, X posts through the fxtwitter API, press interviews, podcasts and conference talks; every quote checked on its page · 140 statements added; verified verbatim by an independent pass
• Event-targeted sweeps (2026-09-14): early dated claims on events already in the registry, found by search on his name and the event; every quote checked on its page by an independent pass · 18 statements added over sweeps 3 and 4; 1 admitted

Ad hoc collection from public statements (MAPS bulletins, conference talks, podcasts and press with direct quotes). Rows are shown and counted; no headline number until the declared tier-B corpus (MAPS bulletins and email updates, verbatim podcast and conference transcripts, MAPS and Lykos releases) is read in full.

• MAPS bulletins and From the Desk of Rick Doblin
• Conference talks and podcasts with verbatim transcripts
• Press interviews with direct quotes
• Second sweep (2026-09-14): MAPS bulletins 2016 to 2026, podcast and video transcripts, press interviews, conference talks; every quote checked on its page · 209 statements added; verified verbatim by an independent pass

The charts follow the lieflat-charts design language (paper, ink, one accent, hairlines, one mark per record), written as our own SVG components; no template code is reused.