Every claim a public programme makes rests on a record somewhere: an agreement that constituted it, a dataset behind a figure, a review that cleared a report, a decision that corrected an error. When those records are missing, scattered or silently changed, three failures follow. Readers cannot tell an estimate from an observation. A programme announced as active may never have been constituted. And a correction made in one place does not reach the copies, citations and summaries that have already spread elsewhere.
Records science, the shared ground of archivists, librarians, data engineers and auditors, treats these as structural problems rather than lapses of care. A record is trustworthy when its identity is stable, its origin can be traced, its integrity can be verified and its history is kept. Failures cascade because records are linked: a superseded dataset quietly feeds a model, the model feeds a report, and the report feeds a decision. Without lineage, no one can find which downstream records an error has touched.
Consortium programmes add a further difficulty. Work is carried out by many institutions across six regions, each with its own authority over what it produces. A shared registry cannot override that authority. It can only make each owner’s records consistent, citable and honest about their status.