Recognises a derived or copied column without a check
A column carrying derivation — a stored total, a copied name, a path string — with no check comparing it with a fresh computation. The update anomaly was accepted on purpose; the check is what makes it safe. A snapshot meant to keep the value as it was is not a copy and needs no check: say so in the derivation.
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 | 64% | 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 | 16% | Not shown |
| gpt-6-luna | Low cost | 31% | Not shown |
| claude-haiku-5-5 | Low cost | 56% | 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 column and its recomputation check
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: account.holder_name copied from party.name with derivation: "copied by ETL"; the ETL step is disabled in a refactor and nobody notices for a year Acceptable: DQ-021, dimension: consistency, scope: set, comparing account.holder_name with party.name through holder_party_id, backs: [PE-057] (DQR-019)
Also known as: Derived or copied column nobody watches (data quality rule)
Related knobits
Builds on
- K-603 Recognises a stored derived or summary value
- K-601 Recognises an unmanaged duplicate column
- K-565 Explains derived data and point-in-time snapshots
All 30 knobits of Derive data quality rules.
Source: Review checklist 03-dq-rules (data-quality), row 4; Kimball & Caserta 2004; this skill (06-physical.md); formerly data-modelling K-704