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).