The method

How I work

Every claim in this investigation is tied to a specific document, and every document is weighed before it is used. The method is simple to state and slow to practice.

The tier system

I rate a source by its provenance before I rely on it, and the rating stays attached to the claim. What decides the tier is not how old a document is. It is where it came from: firsthand or secondhand, a primary record or a later retelling, sourced or merely asserted.

Tier 1: primary and firsthandA document that is itself the record. Written by the people doing the work, at the time, for their own use, not to persuade a later audience. Declassified program files (memos, reports, evaluations, cables), contemporaneous official records, and primary documents entered into the public record, including material submitted to Congress. The test is provenance, not age: a primary record is primary whether it was written in 1975 or last year. This is the load-bearing evidence. A strong claim rests on Tier 1 alone.

Tier 2: secondary, or sourced but uncorroboratedThe journalism of the period, later histories and analyses, and firsthand accounts I can document but have not yet matched against the record. It can corroborate a Tier 1 finding. It cannot carry one by itself.

Tier 3: later accounts and self-reportMemoirs, interviews, and recollections produced after the fact, including by participants, and single sources I cannot corroborate. Useful for leads and context, treated as the weakest evidence, never allowed to override a document.

Where a later account conflicts with the primary record, the record wins. In more than one case the primary documents contradict a participant’s own retelling.

I built a searchable archive of the record

The declassified corpus is enormous and scattered, so the first work was to make it searchable. I download the original files, or photograph physical documents, and run each through optical character recognition to a high-fidelity text transcript. Those transcripts are searchable as plain text and are indexed into a local retrieval system that can cross-reference more than thirteen thousand documents at once, all of it filed and cross-linked in a structured research vault. The point is not the tooling. The point is that a single memo can be found, and set against every other document that bears on it, in seconds instead of weeks.

The search finds the lead. The document decides.

A search hit, a transcript, or a retrieval result is a lead, not a fact. Nothing ships on a transcript. Before a claim is published, the exact words are confirmed against the original page itself, the scan and not a summary of it. This is the safeguard that makes the rest trustworthy: the tools surface candidates fast across the whole corpus, and the primary document, read directly, settles what it says. It is slower work. It is the difference between what a document is said to say and what it says.

I am willing to be wrong in public

Promising and proven are not the same thing. Where the evidence is strong I say so. Where a claim is unproven I say that too. When something I have published turns out to be wrong, I correct it in the open. This investigation is willing to publish findings that are unfavorable to remote viewing, because it is sourced to the record and not to belief.

The method, practiced

The coordinate protocol these documents describe is a defined, repeatable procedure. If you want to run it yourself, honestly and without the mysticism, that is what Scanate trains.

Follow along

New documents and dispatches as they publish. The newsletter is the one place to keep up.