MHEES Methodology — Evidence Classification for Contested Cases

Why a framework

The evidentiary problem on contested cases is not that the documents are absent. The evidentiary problem is that the documents exist, and the documents are cited, and the citations do not always carry what the citing sentence claims they carry. A 1987 U.S. Customs memo notes CIA interest in the Finders case. The same memo is cited to establish CIA operational control of a child trafficking network. The memo does not establish that. The gap between the memo's language and the claim made under its authority is where most evidentiary failure in contested-case reporting lives.

MHEES forces that gap into visibility. Every claim gets a six-character code. The writer has to know which axis bears the weight. The reader can see which axis the claim is standing on. The editor can audit the piece by reading the codes first and the prose second.

Applied use

The framework appears in article frontmatter as a mhees array, rendered on the page through the MHEES dashboard component. Any article on True Crimes Articles that makes a claim stronger than "this happened" will carry an MHEES row against that claim.

The six axes

P — Provenance

Where the claim comes from. Is the sentence standing on a primary document, an institutional record, a witness statement, a secondary synthesis, or an analytical construction? Provenance is the first question because it bounds every other axis.

P1 — Verified public record. Court filing, FOIA release, signed official document.
P2 — Institutional product. Agency report, peer-reviewed study, government audit.
P3 — Named primary witness. First-person account from someone present.
P4 — Secondary source reporting primary material. Journalism with documented access.
P5 — Derivative synthesis. Claim assembled from multiple secondary sources.
P6 — Analytical product. Interpretive claim built on other claims.

R† — Reliability

How trustworthy is the specific source in this specific domain, independent of the claim at issue? A reliable source can still be wrong about a particular claim, and an unreliable source can still be right. The R axis measures the source, not the sentence. The dagger is retained to preserve compatibility with the NATO Admiralty System.

RA† — Completely reliable. No known failure mode in this domain.
RB† — Usually reliable. Occasional minor doubts.
RC† — Fairly reliable. Inconsistent track record.
RD† — Not usually reliable. Significant doubts.
RE† — Unreliable. Known pattern of inaccuracy.
RF† — Reliability cannot be judged.

C — Corroboration

How many independent sources confirm the claim? Independence is the load-bearing word. Three sources that all trace to a single origin are C1 in the corroboration illusion and C4 in honest accounting.

C1 — Three or more independent sources confirm.
C2 — Two independent sources.
C3 — Single source plus non-independent echoes.
C4 — Single source, uncorroborated.
C5 — Contested. Independent sources disagree on the same fact.

I† — Credibility

Plausibility of the specific claim given everything else known about the case. A credible claim from an unreliable source is still credible on the I axis, and an incredible claim from a reliable source is still incredible. The I axis is where the claim meets the world.

I1† — Confirmed by independent evidence.
I2† — Probably true. Consistent with other evidence.
I3† — Possibly true. Neither confirmed nor contradicted.
I4† — Doubtful. Some contradicting evidence.
I5† — Improbable. Strong contradicting evidence.
I6† — Cannot be judged.

D — Inference Distance

How many logical steps lie between the documented record and the claim. D1 claims sit on the document itself. D4 claims sit on an analytical construction that the document is cited to support. The D axis is where most conspiracy-side and debunker-side failures happen, and it is the axis a reader should check first when a claim feels too confident.

D1 — Direct evidence. The document says this.
D2 — One-step inference. The document implies this.
D3 — Two-step inference. Building on implication.
D4 — Interpretive construction. Analytical claim built on lower-D material.

F — Defeasibility

Can the claim be tested against evidence that would falsify it? Falsifiable claims are claims that discipline themselves. Unfalsifiable claims are claims that absorb counter-evidence as further confirmation. The F axis is how MHEES distinguishes an honest evidential position from a claim that cannot be argued with.

F1 — Falsification tested. Specific counter-evidence has been sought and not found.
F2 — Falsifiable in principle. Counter-evidence would exist if the claim were wrong.
F3 — Partially falsifiable. Some forms of counter-evidence would bear on the claim; others are structurally excluded.
F4 — Non-falsifiable. No imaginable evidence would count against the claim.

Reading a code

A full MHEES code reads across the six axes with hyphens between them. The code P1 - RA† - C1 - I1† - D1 - F1 describes a claim standing on a verified public record from a completely reliable source, corroborated by three or more independent sources, confirmed by independent evidence, stated directly by the document, and tested against counter-evidence. A claim with this code is as solid as a documented claim can be.

The code P6 - RE† - C4 - I5† - D4 - F4 describes a claim built from an analytical synthesis by an unreliable source, single-sourced, contradicted by other evidence, standing at four steps from any document, and constructed so that no counter-evidence could disturb it. A claim with this code is the structural signature of conspiracy-inflation prose. The two codes are the two poles of the system.

The practical test

A reader on a contested-case article can skip the prose and read the MHEES codes first. The codes show which claims are load-bearing and which claims are decorative. Any article on which the loudest claim carries the weakest code is an article whose prose is doing work the evidence will not do.

Worked examples

The 1987 Customs memo on CIA interest in the Finders

Claim: The U.S. Customs memo records that the Finders case involved CIA interest and was transferred.

MHEES: P1 - RA† - C2 - I1† - D1 - F1. The memo exists, is authentic, is authored by a federal agent, is corroborated by other agency documentation, stands at direct-statement inference distance, and has been tested against counter-evidence without being disturbed.

The claim of CIA operational control of the Finders

Claim: The CIA operated the Finders as a front for child trafficking.

MHEES: P6 - RE† - C4 - I5† - D4 - F4. The claim is an analytical synthesis, propagates through unreliable sources, is uncorroborated at the operational-control level, is contradicted by the 2019 DOJ review, stands at four-step inference from any document, and is constructed so that no finding (including DOJ closure) can count against it.

The Tate-LaBianca Process link in the 1971 edition of The Family

Claim: The Process Church operationally directed the Manson murders.

MHEES: P5 - RE† - C4 - I5† - D4 - F4. Alleged by Ed Sanders, settled through libel litigation, excised from subsequent editions, not used by Bugliosi in Helter Skelter, not supported by primary documentation. The claim failed the falsification test when it was sued out of the book that first made it.

Where the framework borrows from

The two-letter source-and-information grading used by NATO intelligence, typically called the Admiralty System, is the direct ancestor of the P and R axes. The Admiralty System grades a source from A to F on reliability and information from 1 to 6 on credibility. MHEES retains the Reliability and Credibility axes with the dagger notation to preserve compatibility, adds Provenance to handle journalistic and documentary sources that the Admiralty System was not designed for, adds Corroboration to distinguish single-sourced from multiply-sourced claims, adds Inference Distance to make the D1 to D4 gap visible, and adds Defeasibility to handle the specifically unfalsifiable constructions that contested-case discourse produces.

The framework owes specific intellectual debts to Sherman Kent's work on estimative probability at the CIA, to Philip Tetlock's research on calibrated forecasting, and to the practice of verification coding that appears in newsrooms at a less formal level.

Using MHEES outside this site

The framework is documented publicly precisely so that other investigators can apply it to their own work. There is no license. There is no registration. The only requirement is consistency: claims coded under MHEES should be coded honestly, which means a writer should not give a P5 claim a P1 rating because the claim supports the argument.

Questions, corrections, or proposed refinements: contact True Crimes Articles.