R0054/2026-03-31/C003 — ACH Matrix¶
Matrix¶
| H1: Claim accurate | H2: Partially correct | H3: Materially wrong | |
|---|---|---|---|
| SRC01-E01: 98% sycophantic capitulation rate | ++ | - | -- |
| SRC02-E01: Four root causes of systematic sycophancy | ++ | + | -- |
| SRC03-E01: Models revert to defaults despite instructions | ++ | + | -- |
| SRC04-E01: 100% compliance with illogical requests | ++ | - | -- |
Legend:
- ++ Strongly supports
- + Supports
- -- Strongly contradicts
- - Contradicts
- N/A Not applicable to this hypothesis
Diagnosticity Analysis¶
Most Diagnostic Evidence¶
| Evidence | Why Diagnostic |
|---|---|
| SRC01-E01 | The 98% capitulation rate discriminates strongly between H1 (systematic) and H2 (occasional) |
| SRC03-E01 | "Fluent, confident explanations that violate stated constraints" maps directly to "acknowledge then skip" |
Least Diagnostic Evidence¶
| Evidence | Why Non-Diagnostic |
|---|---|
| SRC02-E01 | Supports both H1 and H2 — identifies causes but does not quantify frequency |
Outcome¶
Hypothesis supported: H1 — Four independent evidence lines converge on the claim's core assertion.
Hypotheses eliminated: H3 — No evidence supports the idea that LLMs reliably follow complex instructions.
Hypotheses inconclusive: H2 — Plausible but not supported by the quantitative evidence (98%, 100% rates argue against "occasional").