Recognises a check stronger than its rule
A check that fails rows the rule allows: a pattern stricter than the format, a closed interval where the rule's is half-open, a value list frozen at an older version. It reports defects that are not defects, and with on_failure: reject it removes valid data.
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 | 80% | Not shown |
| gemini-3.1-pro-preview* | Flagship | 100% | Mastered |
| gpt-6-astra* | Flagship | 100% | Mastered |
| claude-fable-5-1* | Flagship | 83% | Not shown |
| gemini-3.8-flash* | Mid tier | 100% | Mastered |
| gpt-6.1-sol* | Mid tier | 100% | Mastered |
| claude-sonnet-5-5* | Mid tier | 100% | Mastered |
| gemini-3.5-flash-lite | Low cost | 49% | Not shown |
| gpt-6-luna | Low cost | 83% | Not shown |
| claude-haiku-5-5 | Low cost | 82% | 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 the rows it wrongly fails
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: account_status_code IN ('OPN','DOR','CLS') frozen in the check, while the value set now includes BLK; every blocked account is rejected Acceptable: The check reads ref_account_status as at the measurement date instead of a frozen list; the rule owns the value set (row 15)
Also known as: Check stronger than its rule (data quality rule)
Related knobits
Builds on
All 30 knobits of Derive data quality rules.
Source: Review checklist 03-dq-rules (data-quality), row 12; Olson 2003; Loshin 2011; formerly data-modelling K-712