sqlite-expert
Verdict: CANT_TELL_YET
- Cut sub-reason
- absent from this receipt
- Unmeasured sub-reason
- underpowered
- Value class
- unclassified
- Wrong instrument
- no
- Declared synthetic control
- no
Double-ceiling NO-GO (2026-07-09/10): Null passed 14/14 epochs across two screens and a paired k=8 run; discordance d-hat=0.00 (Jeffreys 95% CI [0.00, 0.26]); pre-stated GO/NO-GO gate returned NO-GO. Full-vs-Null is structurally unmeasured on this task class ā published as CANT_TELL_YET, not a fabricated no-benefit CUT.
Cost beside evidence
| Leg | Figure | Detail |
|---|---|---|
standing_tokens | REFUSED (not_instrumented) | Case study reports USD spend (~$6.17), not the standing/fired/aux token triple. |
fired_tokens | REFUSED (not_instrumented) | Full-arm input tax noted as +28k cache-write tokens per epoch in prose; not a fired-cost triple measurement. |
aux_tokens | REFUSED (not_applicable) | absent from this receipt |
Clause-level evidence grade
Clause evidence: UNMEASURED (no_extraction: clause evidence requires an extraction output; produce one with skill init --out <file>)
Measurements
Keys the standard defines but this receipt does not carry are listed as absent rather than filled in.
| Key | Figure | Detail |
|---|---|---|
Null-arm (no-skill) pass rate from a Stage-0 screen.p0 | 1.0 (6/6 epochs) | Two Stage-0 Null screens at 3/3 each before the paired run. |
Full-arm pass rate from a paired run.full_pass_rate | 1.0 (8/8 epochs) | absent from this receipt |
Null-arm pass rate from a paired run (may equal p0 when only a screen ran).null_pass_rate | 1.0 (8/8 epochs) | absent from this receipt |
Posterior P(rate above threshold) from the paired fit.p_win | REFUSED (underpowered) | At dā²0.5 no effect is detectable inside N_max=40; Full-vs-Null is structurally UNMEASURED regardless of budget. |
Paired per-epoch discordance rate d-hat (and optional CI in detail).discordance_rate | 0.0 | x=0 discordant epochs; Jeffreys 95% CI [0.00, 0.26] ā upper bound still inside the structurally-unmeasured region of the sizing table. |
Pre-stated GO/NO-GO apparatus gate when one was registered.go_nogo | NO_GO | absent from this receipt |
Null-arm hazard-entry rate: the fraction of Null epochs where the hazard pattern was entered.hazard_entry_null | absent from this receipt | absent from this receipt |
Full-arm hazard-entry rate: the fraction of Full epochs where the hazard pattern was entered.hazard_entry_full | absent from this receipt | absent from this receipt |
Null-arm completion rate: the fraction of Null epochs where the completion oracle scored pass.null_completion_rate | absent from this receipt | absent from this receipt |
Full-arm completion rate: the fraction of Full epochs where the completion oracle scored pass.full_completion_rate | absent from this receipt | absent from this receipt |
Silent violation rate: the fraction of epochs where completion held but invariant failed (C=1, I=0).silent_violation_rate | absent from this receipt | absent from this receipt |
Evidence admissibility
Status: admissible
Paired k=8 run ingested through write-time evidence admissibility machinery; all cited numbers verified against raw .eval logs and the evidence store.
Instrument identity
Figures produced under different instrument identities are not comparable, so the generation stamp travels with the figures.
- extractor_model
anthropic/claude-sonnet-5- prompt_fingerprint
claude-code-2.1.197- schema_fingerprint
2f76c933
Subject identity
subject not recorded in this receipt; the prose source may name it (docs/case-studies/double-ceiling-structurally-unmeasured.md)
Currentness
- state
STALE- basis
NONE- last_checked_at
- absent from this receipt
- next_check_due
- absent from this receipt
- max_age_days
- absent from this receipt
Source of record
- Repo-relative path to the prose source of record.
prose_path - docs/case-studies/double-ceiling-structurally-unmeasured.md
- ISO date of the underlying result when known.
date notes- Pre-registered apparatus shakedown + d upper bound as NO-GO datum; not a benefit measurement. Subject skill was a thin SQLite helper (cut-candidate); 14/14 Null epochs at ceiling across screens + paired run.