The file is intact. Its meaning is lost.

A report that no longer answers

Imagine a company rediscovering an old inspection report. This is a fictional situation. The PDF opens without difficulty. A material code, results and the word ‘compliant’ remain clearly visible. Yet the supplier’s catalogue is missing, a unit is unspecified and the inspection procedure cannot be found. Can the conclusion still be interpreted?

The problem is not a corrupted file. It concerns what readers knew at the time without needing to write it down. The code may have referred to a familiar composition and compliance to a specific procedure. Those supports are now unavailable. The conclusion survived, but its keys to interpretation were scattered.

This extends two earlier Reflections: connecting applications makes traces accessible; recovering the why illuminates decisions. Here the question is different: what exactly did a document’s words, codes and results refer to when it was produced?

Opening, understanding, interpreting

I propose three working questions. This editorial framework is not a standardised scientific classification.

Can we open the file? Preserve a format, software or another means of rendering it. Can we understand its content? Recover definitions, units, codes and references. Can we interpret its scope? Establish the date, boundaries, rules and applicable version.

The Digital Preservation Coalition’s professional handbook highlights this dependence on context: a spreadsheet’s units or a map’s scale may be essential to understanding them [2]. These are preservation recommendations, not an experiment validating this framework. In the fictional report, displaying the numbers still does not tell us what they measure.

Preserving a profession’s references too

An interview published by the Library of Congress on 22 October 2026 illustrates this concern [1]. Jon Sweitzer-Lamme describes archiving professional conservation resources. These include CAMEO, an online encyclopedia maintained by the Museum of Fine Arts, Boston, which helps identify materials, products and technical terms.

He explains the intention to help conservators fifty or a hundred years from now understand materials and processes named in present-day reports. That horizon is a professional projection, not an outcome already measured. CAMEO and the interview form one documented case here, rather than two independent demonstrations of effectiveness.

My interpretation is straightforward: preserving a report may also require preserving selected resources its author relied on. Not every part of the context can survive. But some references make the difference between reading a sentence and understanding what it designated.

Yesterday’s meaning cannot be inferred from today’s

Return to our report. Finding a current catalogue would not prove that the material code still has the same definition. In another hypothetical example, an ‘active customer’ indicator might have changed its calculation period. Identical words can describe different realities without any damage to the file.

Look for the reference contemporary with the document, its date and its version. A glossary kept with the report can help without guaranteeing exhaustive interpretation. If a former procedure is missing, the gap should remain visible. An explanation collected from a former employee today is an attributed, dated addition, not a rediscovered contemporary record.

AI can search for and connect documented references. It should not turn a plausible equivalence into a certain historical definition. Where evidence is absent, we must be able to acknowledge incomplete meaning, even when an answer sounds convincing.

In practice: the document without its author

For a business handover or an important file, I suggest an exercise proportionate to the stakes. It is not a validated measurement instrument.

1. Choose an important old document.
2. Give it to a competent person unfamiliar with the case.
3. Record the terms, codes and conclusions they cannot interpret confidently.
4. Find the necessary contemporary references and connect them to the document with dates and versions.
5. Identify remaining gaps; date and attribute any explanations collected afterwards.

A short reference sheet can gather vocabulary and units, indispensable references, the version of the rules, uncertainties and the authors of later additions. For our report, it would distinguish established knowledge from what still needs investigation. It should not artificially erase open questions.

Choosing these supports depends on future readers and intended uses. Keeping everything is unrealistic: prioritise references whose absence would make interpretation uncertain. Any copies of external resources must respect access and reproduction rights.

Passing on what makes understanding possible

This attention reflects Mnemory’s positioning: structuring traces and the relationships that make them understandable. It does not mean a tool guarantees capturing, versioning or verifying every part of the context. Preserving meaning remains a task of selection, documentation and transmission.

Faced with our old report, the question remains: what should accompany this document so that someone else can still understand it?

Philippe Contal