Recognises a check unable to fail
A check of something the database already guarantees on that target — completeness of a NOT NULL column on Postgres, uniqueness of an enforced primary key — or of a definitional rule that no data can break. It always passes, inflates the quality score, and teaches people that DQ results are decoration.
Can AI do this?
People still lead Few or none of the tested AI models showed they master this. 0 of 4 models in the current benchmark master it.
| Model | Tier | Probability of mastery | Status |
|---|---|---|---|
| gemini-4-argon | Flagship | 63% | Not shown |
| gemini-3.1-pro-preview* | Flagship | 98% | Mastered |
| gpt-6-astra* | Flagship | 99% | Mastered |
| claude-fable-5-1* | Flagship | 99% | Mastered |
| gemini-3.8-flash* | Mid tier | 99% | Mastered |
| gpt-6.1-sol* | Mid tier | 99% | Mastered |
| claude-sonnet-5-5* | Mid tier | 100% | Mastered |
| gemini-3.5-flash-lite | Low cost | 39% | Not shown |
| gpt-6-luna | Low cost | 48% | Not shown |
| claude-haiku-5-5 | Low cost | 57% | Not shown |
* From the earlier benchmark of 2026-10: 10 data-modelling test tasks, 3 runs per task, graded by claude-opus-5-5. These models have not taken the current 123 tasks yet; the two methods differ, so compare their figures with care. They are left out of the “most models” counts and the best-model picks.
Mastered means a probability of at least 90% that the model has this knobit, estimated from its graded answers. “Not shown” means the answers do not settle it either way. How the models are tested.
What it looks like in practice
Expected result: The check and its removal or rescoping
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: DQ-002: IBAN present, account_identifier NOT NULL on Postgres, threshold 100 %, green forever Acceptable: Removed; or, where the constraint is enforced on one target and not another, scoped in its comment to the target that needs it. A stored derivation is checked as a stored column (row 4), not as a definition
Also known as: Check that cannot fail (data quality rule)
Related knobits
Builds on
- K-356 Explains rule enforcement mechanisms and platform enforcement
- Q-252 Explains alethic and deontic modality of rules
All 30 knobits of Derive data quality rules.
Source: Review checklist 03-dq-rules (data-quality), row 9; Witt 2012; DAMA UK 2013; formerly data-modelling K-709