Skip to content

R0044/2026-03-29/Q004 — ACH Matrix

Matrix

H1: Both system-side and human-side H2: Only human-side H3: System properties but not output behavior
SRC01-E01: CaTE Guidebook — trust and trustworthiness - + ++
SRC02-E01: SEI/CMU — "human has to understand" -- ++ +
SRC03-E01: Sandia TCMM — communicating trustworthiness -- + ++

Legend: - ++ Strongly supports - + Supports - -- Strongly contradicts - - Contradicts - N/A Not applicable to this hypothesis

Diagnosticity Analysis

Most Diagnostic Evidence

Evidence ID Why Diagnostic
SRC02-E01 Sablon's quote is the most diagnostic — it explicitly frames the problem as human understanding, strongly contradicting H1

Least Diagnostic Evidence

Evidence ID Why Non-Diagnostic
SRC01-E01 The Guidebook description is general enough to support both H2 and H3

Outcome

Hypothesis supported: H3 — CaTE addresses system design properties (trustworthiness dimensions) but not system output behavior (sycophancy, trust-adaptive output).

Hypotheses eliminated: H1 — No evidence of system output behavior in CaTE's scope.

Hypotheses inconclusive: H2 — Partially supported (human-side is primary focus) but too narrow (CaTE does address system properties).