A municipal sapling-yard helper’s wrong hardiness zone can be walked back to one donated spreadsheet from 2014. What control would have named that source, date, and transform?
Select an answer to reveal the explanation.
Short Explanation
Think of a wrong hardiness zone that walks back to one donated 2014 spreadsheet. Lineage would have named the source, date, and transform. A font change does not.
Full Explanation
When a municipal sapling-yard helper's wrong hardiness zone walks back to one donated spreadsheet from 2014, the missing control is data lineage for what trained or grounded the system. Lineage names the source, date, and transforms so operators can find, quarantine, and correct bad inputs instead of endlessly tuning prompts around mysterious errors.
A font change fails because typeface styling has nothing to do with whether a 2014 spreadsheet poisoned hardiness zones. Prompt caching fails because reused static prefixes optimize inference; they do not record corpus provenance. From-scratch pre-training fails because launching a weekend crawl does not create a lineage record for an existing donated file already in the pipeline.
Exam caveat: lineage is only as good as the discipline of recording transforms—ad-hoc spreadsheet edits without tickets break the trail even if an initial source was logged. Operational check: tag each training or grounding dataset with source URI, ingest date, and transform job ID; for the sapling-yard incident, locate the 2014 spreadsheet via that registry, remove or correct it, and rebuild or re-index so the wrong zone cannot recur from the same artifact.