C001 — ACH Matrix¶


Research	R0058 — Candidate evidence test
Run	2026-04-03
Claim	C001

Matrix¶

	H1: Claim accurate as stated	H2: Partially correct (83% yes, 1% mischaracterized)	H3: Materially wrong
SRC01-E01: 83.1% global homophily confirmed	++	++	--
SRC01-E02: Top 1% by degree control 58% of paths (not "1% bridging")	--	++	N/A
SRC01-E03: 9.5% mixed papers; 16.9% reachability	-	+	+
SRC02-E01: 94% connected component in broader AI ethics	N/A	+	+
SRC03-E01: Safety underrepresented in ethics venues	+	+	--
SRC04-E01: NeurIPS workshop on safety-ethics gap (2019)	+	+	-

Legend:

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

Diagnosticity Analysis¶

Most Diagnostic Evidence¶

Evidence	Why Diagnostic
SRC01-E02	This is the single most diagnostic piece of evidence. It strongly supports H2 while strongly contradicting H1, because it reveals that the claim's "1% of authors bridging" is a mischaracterization of the paper's actual finding about the top 1% by degree controlling 58% of paths. This evidence cleanly separates H1 from H2.
SRC01-E01	Strongly supports both H1 and H2 while strongly contradicting H3. Confirms the 83.1% homophily figure with robust methodology, eliminating the possibility that the core phenomenon is fabricated.

Least Diagnostic Evidence¶

Evidence	Why Non-Diagnostic
SRC02-E01	Measures a different network (countries/institutions vs. individual co-authors) and a different scope (AI ethics broadly vs. safety-ethics specifically). Supports both H2 and H3 without discriminating between them.
SRC04-E01	Qualitative evidence that supports both H1 and H2 equally. The workshop's existence confirms the divide but says nothing about the specific numbers.

Outcome¶

Hypothesis supported: H2 — The claim is partially correct. The 83% homophily figure is confirmed (83.1%), but the "1% of authors bridging" is a mischaracterization of the source's finding about network centrality concentration.

Hypotheses eliminated: H1 — Eliminated because the "1% bridging" component does not accurately represent the source finding. H3 — Eliminated because the 83.1% homophily figure is robustly confirmed by the primary source.

Hypotheses inconclusive: None.