R0049/2026-03-31/Q001 — ACH Matrix¶
Matrix¶
| Evidence | H1 Full framework exists | H2 Nothing exists | H3 Partial only |
|---|---|---|---|
| SRC01-E01 Reporting checklist | N/A | - | ++ |
| SRC02-E01 Framework CoT screening | + | -- | ++ |
| SRC03-E01 Library gap | - | + | ++ |
| SRC04-E01 LLM SATs | + | -- | ++ |
| SRC05-E01 Agent Lab no rigor | - | N/A | ++ |
Legend¶
| Symbol | Meaning |
|---|---|
| ++ | Strongly consistent |
| + | Consistent |
| - | Inconsistent |
| -- | Strongly inconsistent |
| N/A | Not applicable |
Diagnosticity Analysis¶
Most diagnostic evidence:
-
SRC04-E01 — Roberts' SAT implementations are the most diagnostic because they simultaneously provide evidence against H2 (implementations exist), weak evidence for H1 (multiple techniques implemented), and strong evidence for H3 (separate tools, not unified framework). The distinction between "three tools" and "one framework" is the key discriminator.
-
SRC02-E01 — Framework CoT is diagnostic because it shows framework-guided prompting works well (against H2) but covers only screening (for H3, against H1).
Least diagnostic evidence:
- SRC01-E01 — The reporting checklist is tangential to the core question of whether operational system prompts exist.
Outcome¶
H3 (Partial implementations only) is strongly supported by all five evidence items. H2 is eliminated by two items of strongly inconsistent evidence. H1 receives only weak partial support. The ACH matrix clearly favors H3.