Application packet
An upload form takes files, so at some point every one of these had to stop being a web page and become a PDF. That conversion loses the part that matters: in the file the citations are dead text, and you have to take the numbers on trust. So both exist. The PDF is what gets attached; the page is what you can check. They are rendered from the same source on the same run, which is why the page counts below are stated — if a file ever disagrees with its page, that is a bug and I want it visible.
For thirty years, writing about the Harvard psychiatrist John E. Mack has repeated one figure for Laurance Rockefeller's funding of his work: a million dollars, four years, $250,000 a year. The figure cites a source that says two years. And the foundation any researcher would check first to settle it — the Laurance S. Rockefeller Fund — did not exist until two years after Rockefeller was dead. The number is not merely unverified; the trail that would verify it has been pointing at the wrong door. This is the document to read first if you only read one, because it is the only one where the method produces a finding rather than describing itself.
136 episodes read end to end, every factual claim extracted, classed by what kind of evidence would settle it, and scored — 850 entities, twenty-eight anomaly sites against one seventeen-point checklist. The live instrument has since been retired from this site; the printed summary is the record of how it worked and what it found.
One page, and every claim on it that can be checked links to the thing that checks it — including a retraction, which is on the résumé because leaving it off would misrepresent how the work actually went.
Six beats on the Mack funding thread, at length, in narration. Two of them carry findings that are not published anywhere else: the foundation that could not have made the grant, and the death index that settles a thirty-year rumour in two entries. It opens by correcting an inflated credential that flatters its own subject, and it declines to state a dollar figure, because the absence is the finding.
The page shows its working; the PDF does not. Print it and you get the script alone. Read it in the browser and every beat also carries the evidence class behind it, the source facts, and the voice rule it was written to — so a claim in the narration can be traced to the thing it rests on without taking my word for any of it.
Every attachment above has a live page behind it, and this page opens a run that can be read straight through: the finding, the work sample, the résumé. Each page ends with the next one, named — nothing here requires the navigation.
If you would rather start with the person than the work, the overview is a separate and much shorter read.