This is a workbench, not a document. Everything grey is scaffolding I built —
the beats, their evidence class, the source facts, and the voice rule that applies to each one.
The bordered blocks are yours to write in.
Hit Cmd-P at any point and the scaffolding vanishes: the PDF contains only your narration,
typeset as a script. That is the artefact they read. Any beat you have not written prints a red
marker instead of silently shipping a gap.
The full brief — spine, the eighteen measured voice rules, and the traps — is in
02_owner/WRITING_BRIEF_2026-08-05.md. The two original findings are in
03_dossier/L13a_irs_entity_check.md.
Your voice on this one is the whole point; the prose cannot be delegated.
John E. Mack — spec segment
Work sample for American Alchemy · target 1,200–1,800 words, eight to twelve minutes read aloud ·
written in Jesse's voice
The spine. The story everyone tells about Mack is a story about vindication, and nearly every
load-bearing detail of it is either wrong or unverified — including the details his defenders repeat.
Not a debunk: the strongest version of the Mack case has to be rebuilt from the documents,
because the popular version cannot survive them.
ACT I
The credential, corrected
Established — write these as flat statements
- Fact Professor of Psychiatry at Harvard Medical School
from 1972 until his death; tenured throughout.
- Fact 1977 Pulitzer Prize for Biography,
A Prince of Our Disorder.
- Fact Founded and led the Cambridge Hospital psychiatry
department, 1969–1977; held the chairmanship of the Executive Committee.
The correction that opens the piece
"Head of Harvard psychiatry" is not a post that exists — HMS psychiatry is a federation of
hospital departments with no single resident head. Correcting an inflation that
flatters your own subject in the first ninety seconds is what buys the audience's trust
for everything after it. Their readers are described as allergic to being sold to.
Voice
STY-01 open on one of the four hooks — date-stamped scene, famous case
plus twist, blunt question, or a challenge to what the viewer already believes.
STY-02 thesis plus the episode's table of contents inside ninety seconds.
STY-08 no exclamation marks, ever.
ACT II
The investigation
Established
- Fact Dean Tosteson appointed a peer committee; stated
grounds were method and clinical conduct, explicitly not subject matter.
- Fact No charge letter, and no existing university
procedure to follow.
- Fact Chaired by Arnold Relman, editor-in-chief emeritus
of the New England Journal of Medicine.
- Fact ~30 meetings over ~14 months, May 1994 to
3 August 1995. Harvard took no action — and the Dean issued a warning. He did not simply win.
- Reported The BBC states flatly it was the first such
inquiry of a tenured Harvard professor. Relman called it "unprecedented." Harvard refused to
characterise it — so the superlative is reported, not established.
- Fact (negative) No AAUP statement or censure exists.
Checked.
The detail that makes this an act rather than a summary
It became public because Mack's own lawyer circulated a letter soliciting defence
witnesses that quoted the committee's draft report. The defence leaked it.
Voice
STY-04 announce the track switch and ask the viewer to stay.
STY-12 tag the BBC superlative inline as reported, at the moment it appears.
ACT III
The steelman — give the case against him its strongest form
Established
- Fact His own peers said the problem was that he was not
doing science.
- Fact He conceded he was not an expert on hypnosis and
defended it in terms that are not scientific — while insisting publicly and repeatedly that
hypnosis was not his primary route.
- Fact The Donna Bassett sting: fact as to what was said.
Inference what it proves is yours, and must be marked.
Correct the show's own past framing here
In Greer's telling the funder is not Rockefeller; the episode has stated it the other way
round. Correcting a prospective employer, politely and with the receipt, is a stronger signal than
agreeing with them.
Voice
STY-16 give contested evidence an outright verdict on its strength.
Do not turn the steelman into a strawman — the posting asks for both held at once.
ACT IV
The money Original finding
Established
- Fact Rockefeller funded the Center for Psychology and
Social Change, and PEER, from 1993.
- Fact The amount is not established. Three
incompatible figures circulate; the most-cited — $1M over four years at $250,000 — misreports
its own source, which says two years.
- Fact, primary The Laurance S. Rockefeller Fund that files
with the IRS was created in 2006, two years after Rockefeller died, and cannot hold a 1993
grant. EIN 20-4192256, ruling date June 2006.
The line that is actually the finding
Anyone doing the obvious diligence pulls that foundation's filings, finds no Mack grant, and
concludes the funding story is unsupported — a false negative built out of authentic
documents.
Where it stands, stated exactly
Three folder-level entries in the Laurance S. Rockefeller papers at the Rockefeller Archive
Center, Room 5600 General Files, Science series, plus a companion folder titled "UFO – John Mack."
Inference the likely booking channel is the Fund for the
Enhancement of the Human Spirit — press-documented as his personal giving vehicle in those years,
but only for other gifts; nothing ties it to Mack yet.
Hard constraints
Do not write a dollar figure. The absence is the story. Do not claim the archive folders say
anything — they have not been read.
Voice
STY-05 after a striking number, land it with a two-to-five-word hammer.
STY-03 signpost the escalation before the big claim arrives.
STY-17 falsifiability as the moral test — a claim kept liminal for thirty
years is the most Jesse-shaped sentence in this material. Write it.
ACT V
The death Original finding
Established
- Fact Killed by a drunk driver in London,
27 September 2004.
- Fact A rumour circulates that other John Macks died in
England the same day — one in Alberino's telling, three in a stronger version.
- Fact The complete GRO death index for registration year
2004 contains exactly two male John Mack entries in the entire country, registered in May
and in August. A 27 September death cannot be registered in either.
- Fact Our subject is a unique entry: John Edward Mack,
born 4 October 1929, registered June 2005 in Barnet — a nine-month lag, the ordinary shape when
a coroner opens an inquest and criminal proceedings run first.
- Inference that this kills the rumour. Sound, and
must be marked as inference.
Say the residual out loud
The index publishes month of registration, not day of death, so in principle a second man could
have died that day and been registered late too. The only other June 2005 male John Mack is John
Adolphus Mack, born 1913, Kent — wrong cohort, wrong district, and nobody has ever claimed he was
hit by a car. Conceding that is not weakness; STY-14 is literally the
rule that you concede the unknown and pivot to the firmest fact, and it is how he already writes.
CLOSE
Open, not resolved
The question that earns it
The man's defenders and his detractors have spent thirty years arguing about a figure neither
side checked — and the documents that would settle it have been sitting in an open collection in
Sleepy Hollow the whole time.
Voice
STY-06 open-question crescendo, then the sign-off, with any CTA as a
separate block after it. STY-15 where the answer is unknowable, enumerate
the live hypotheses and leave them open rather than resolving.
Traps
- Do not perform certainty you do not have. It is the one failure mode this subject punishes.
- Do not write a dollar figure for the Rockefeller money.
- Do not turn the steelman into a strawman you knock down.
- No exclamation marks.
- Do not let the research swamp the narration. If it reads as a memo, it has failed the first
bullet of the job.
- Do not claim the archive folders say something. They have not been read.
Need this scaffold, as printed as a file — to keep, or to forward to somebody who has no reason to open a stranger's website?PDF · 1 page · 91 KB
Aleksei Beliankin · spec segment for American Alchemy · printed from a workbench, which is why
the reasoning is missing here: each beat's evidence class, its source facts and the voice rule
that governs it are on the live page at ab7.ai/AA/work-sample · the funding thread this
segment narrates is adjudicated in full at ab7.ai/AA/dossier
—