Evidence architecture and provenance · Lesson 6

Prioritising documentation gaps

5 min read

Lesson objectives

  • Prioritise documentation work under limited staff time; produce a risk-ranked backlog with escalation triggers.
Reading tools

When a small team finds ten catalogue gaps, it cannot close them all before an exhibition opens. Ranking them only by ease of repair may leave the most consequential public statement unsupported. The order must connect each gap to a decision depending on it. A gap that could misname a person or release sensitive material deserves attention before an uncertain ornament description, even if the latter is easier to fix. Yet calling every task urgent prevents any progress. Prioritisation is a decision practice: identify the consequence of error, evidence that the error may exist, how it could later be detected, staff time needed to check it, and an action, deadline and owner.

Suppose an institution is preparing to release a collection of 400 photographs. Four reports arrive: (A) twelve images have conflicting names on old cards and a newer catalogue; (B) sixty have inconsistent colour descriptions; (C) one is linked to a documented request by a narrator to review display of their name; (D) thirty use print-stamp dates as exposure dates. Do not give each report an invented numeric score and obey the sum mechanically. For C, suspend display of the affected name and route the request to access review immediately, though it concerns one item. D may distort an entire timeline: temporarily relabel stamps as print dates instead of exposure dates. A needs identity review before the names appear, but the photographs might be shown under unnamed titles. B may wait for an editorial cycle if it does not affect retrieval or material understanding.

Make a priority table that keeps impact separate from effort. For each task record the dependent decision, harm if left unresolved, evidence that a problem exists, a mitigation available now, a test that would close it, estimated team time and review date. Use explicit categories: “hold a specific output temporarily,” “repair before launch,” “release with qualified wording and follow-up,” and “defer improvement.” Do not hide a secret scoring system behind them. When potential harm is high but the indication is weak, it can be reasonable to restrict only the affected output pending review, without describing the weak indication as proof. If a gap cannot be closed before launch, state the boundary of the public description and who will receive a later report.

When estimating work, separate quick mitigation from root investigation. Removing an unverified name from one page can take minutes, but resolving identity could require interviews, permissions and questions outside the cataloguing team's predictable time. Make two linked tasks: “mitigate before launch” and “investigate after launch with a review date.” Do not allow a temporary state to become permanent by neglect. If a gap concerns a legal right or a community request, specify an escalation point to the appropriate authority instead of letting a cataloguer invent a ruling. Referral does not stop tasks that do not depend on that decision.

Test the ranking by changing one fact. If a photographer's register establishes that the stamps on all thirty images are indeed exposure dates, D drops in priority; C's name-review request does not. If inconsistent colour descriptions turn out to distinguish two versions of a historically important object, B rises from cosmetic improvement to an identity problem. Priority follows the affected decision rather than the field label. Record why it changed so the team's revision appears reasoned, not arbitrary.

Calculate capacity before making promises. In the ten-hour exercise below, restricting the name takes half an hour, qualifying the dates one hour, and reviewing public names four hours: five and a half hours in total. Four and a half remain. The team might start a divisible check of stamped photographs, but it must not mark the investigation of all ten as complete before the estimated five hours have actually been used. Alternatively, if frame descriptions affect retrieval, two hours could go there and two and a half to an initial stamp check with a dated follow-up. “Investigation started” is not “exposure dates established.” Reserve some time to record evidence and update the page; a schedule filled to the minute can itself create documentation mistakes.

Consider the cost of withholding everything. Blocking an entire collection because of one image deprives people of material that can be described safely, while ignoring the request harms the person concerned. Constrain a temporary hold to the relevant file or field where feasible, and document why the restriction does not automatically cover the rest. Expand the review only if new evidence warrants it, with a recorded reason.

Lesson summary

When a small team finds ten catalogue gaps, it cannot close them all before an exhibition opens