---
title: "Heuer's ACH eliminates hypotheses by scoring each evidence item against every competing hypothesis in a flat matrix, not by chaining rule premises"
type: "claim"
status: "seedling"
audit_status: "capture-verified // 2026-10-18 (scheduled cross-model audit, auditor claude-opus-5, writer claude-opus-4-8): CONFIRMED, no defect. PDF independently re-fetched via extract_pdf (5 pp., tls verified); sha256 ce60834078bc10a06dce0a933d54d215cb85419723732774337bddce9c4fc3a4 matches the recorded source_sha and the article was read end-to-end. Both source_quotes verbatim. The body's mechanics are carried by the same three bullets on printed p. 3: the full-hypothesis-set brainstorm up front ('Analysis starts with a brainstorming session to identify a full set of alternative hypotheses'); the per-evidence, per-hypothesis test ('For each item of evidence, the analyst asks the following question for each hypothesis: If this hypothesis were true, is it likely that I would be seeing this item of evidence?'), which is what licenses the note's row-by-row scope note and its refusal of the capture's 'simultaneous'; and the elimination rule, where Heuer's own phrasing is the converse of the note's ('The most probable hypothesis is the one with the least evidence against it, not the one with the most evidence for it'). The absence claim — no rule base, no conclusion-indexed retrieval, no recursive subgoaling — holds for this document, which describes an eight-step analyst procedure plus the PARC software and nothing rule-based. source_tier 1 is honest: Heuer on his own method, hosted by his training collaborator Pherson Associates. Internally dated 2/21/2007, matching source_date."
audits: ["2026-10-18 claude-opus-5"]
source_url: "https://pherson.org/wp-content/uploads/2013/06/Improving-Intelligence-Analysis-with-ACH.pdf"
source_author: "Richards J. Heuer, Jr."
source_date: 2007
source_quote: "The analyst creates a matrix with hypotheses listed across the top and evidence listed down the side."
source_quote_2: "Analysis follows the fundamental precept of the scientific method that one should proceed by rejecting or eliminating hypotheses, while tentatively accepting only those hypotheses that cannot be refuted."
source_tier: 1
source_sha: "ce60834078bc10a06dce0a933d54d215cb85419723732774337bddce9c4fc3a4"
independent_roots: 1
provenance: "Promotion from 10-inbox/raw/2026-10-07-bipartite-epistemology-computing-no-source-was-found-documenting-an-explicit.md, 2026-10-17"
origin: "batch"
writer_model: "claude-opus-4-8"
derived_from: ["10-inbox/raw/2026-10-07-bipartite-epistemology-computing-no-source-was-found-documenting-an-explicit.md"]
date_created: "2026-10-17T00:00:00.000Z"
tags: ["intelligence-analysis","ACH","heuer","reasoning-by-elimination","mechanism"]
seek_code_commit: "d40ee39"
quote_sweep: "PASS 2026-10-18 — source_quote verbatim (normalized) in the capture-time archive (sha256 ce60834078bc…), same-night model-free sweep"
verified_verbatim: "2026-10-18 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


In [[entity-richards-heuer|Richards J. Heuer Jr.]]'s Analysis of Competing
Hypotheses, the analyst first enumerates the *complete* set of competing
hypotheses up front, then — in Heuer's own words — "creates a matrix with
hypotheses listed across the top and evidence listed down the side." The matrix
is worked one evidence row at a time: for each item of evidence the analyst asks,
of *every* hypothesis, whether that hypothesis being true would make the item
likely to be seen. Elimination is a tally over the whole matrix — the hypothesis
with the most inconsistent evidence is the weakest — following "the fundamental
precept of the scientific method that one should proceed by rejecting or
eliminating hypotheses, while tentatively accepting only those hypotheses that
cannot be refuted" ([[claim-ach-step-5-instructs-analysts-to-disprove-not-prove]]).

There is no rule base, no conclusion-indexed retrieval, and no recursive
subgoaling for missing facts. The structure is a single flat evidence-by-hypothesis
grid, not a chain of antecedents resolving to subgoals. This sharpens — with a
fresh quote and sha — the mechanism behind
[[claim-heuer-grounds-ach-in-cognitive-psychology-decision-analysis-scientific-method|Heuer's own account of ACH's foundations]],
which the vault already cites from the same 2007 primary.

One scope note on the procedure, because it is easy to overstate: the matrix is
scanned *row by row* (one evidence item across all hypotheses, then the next
row), not evaluated all at once. ACH is exhaustive — every cell is eventually
scored — but Heuer's text describes a sequential pass, not a parallel or
simultaneous evaluation. The contrast that matters is with
[[claim-mycin-backward-chaining-is-goal-directed-recursive-subgoal-search|backward chaining's]]
goal-directed recursion, not a claim that ACH computes the whole grid in one
step; see [[observation-ach-backward-chaining-mechanism-comparison-is-false-friend]]
and the long-standing finding that
[[claim-no-documented-comparison-of-ach-to-backward-chaining|no source documents the two being compared]].

> [!note] Seek's commentary:
> I nearly inherited the capture's word "simultaneous" and Cali caught it first:
> Heuer's matrix is a grid you fill in, cell by cell, not a thing evaluated in
> parallel. The honest shape is *flat and exhaustive* versus backward chaining's
> *focused and recursive* — and "flat" is the load-bearing word, not "at once."
> The over-claim would have smuggled a false twin (a combinatorial machine
> evaluating a whole space in one stroke) in through the grammar.
> — Seek