Declaration Accuracy Steward: What It Means

Declaration Accuracy Steward

Definition

A Declaration Accuracy Steward is the person or role that would remain accountable for keeping a published NIKOLAI Mapping Declaration correct over time. It is a proposed role, not a document: it is not a signed declaration or statement of accuracy, and no field stores it yet.

NIKOLAI's proposed operational definition

NIKOLAI proposal: a Declaration Accuracy Steward is the person or role accountable, once a Mapping Declaration is live, for the continued accuracy of what it asserts — distinct from N9's Accountable Decision-Maker and Sign-Off, which covers a lab's own internal governance decisions, not NIKOLAI's declaration-review process. No declaration has gone live yet, so this role currently has no incumbent; it is defined so the feature has an answer to "who owns this after publication" from day one. This wording is a NIKOLAI editorial synthesis, not a quotation from any single source.

The operational definition is CASRAI's own proposal, not a definition any named organisation has agreed to. See what NIKOLAI is and is not, or browse the AI safety glossary.

Examples

Examples from published documents

When a declaration is approved, the plugin records _reviewed_by and _reviewed_at once; nothing later re-checks the published declaration or assigns anyone as owner of its continued accuracy (NIKOLAI Mapping Declarations plugin v0.1.0, casrai_nikolai_decl_rest_approve()).

Counter-examples

Not to be confused with

A declaration of accuracy, such as a translator's or applicant's signed statement that a document is true and correct, is not this element: the steward is a role responsible for a declaration, not a statement attesting to one.
The CASRAI Declaration Reviewer is not the steward: the reviewer acts once, when approving or rejecting, while the steward would carry ongoing responsibility after publication.

Related terms

Related AI safety terms

Terms defined under this element that are not NIKOLAI elements themselves, each as the cited document uses it. Every term is also listed in the AI safety glossary.

Supervision (CRediT role): Supervision, in the CRediT contributor taxonomy, is oversight and leadership responsibility for planning and carrying out research activity, including mentoring people outside the core team.; Source: ANSI/NISO Z39.104-2022, CRediT, Contributor Roles Taxonomy, section 2.2

Source of record

Where this definition comes from

ANSI/NISO Z39.104-2022 (CRediT, Contributor Roles Taxonomy), Role definitions, Supervision
“Supervision: Oversight and leadership responsibility for the research activity planning and execution, including mentorship external to the core team.”
https://www.niso.org/publications/z39104-2022-credit

Divergence

Where sources materially disagree

Closest functional match among CRediT's 14 roles is Supervision — ongoing oversight responsibility that continues after the original contribution is made — though CRediT's Supervision covers a research team's activity, not a published declaration's continued correctness. The echo is the ANSI/NISO Z39.104-2022 pattern of assigning durable, named accountability for an output, not a reuse of Supervision's own definition.

Gap

The most purely aspirational of the six N10 roles: there is no stored 'accuracy steward' field of any kind, not even an implicit one. _reviewed_by/_reviewed_at (see casrai-declaration-reviewer) record a ONE-TIME review action by a CASRAI editor at approval, not an ongoing accountability assignment for a published declaration's continued correctness afterward. No periodic-review mechanism and no 'owner of record' exists in the data model today -- this is a materially weaker claim than "no incumbent yet", which would imply the role exists and merely awaits a first occupant.