EDUKATE SINGAPORE · GOLD STANDARD PRACTICAL CASE STUDIES · CASE 150
A real signature on an older document does not automatically authenticate the revision people now read.
A university report’s version 4 is digitally signed, then its unsigned version 5 is published with a corrected chart and additional appendix while retaining the old signed badge. This fictional Gold Standard case follows content integrity, signature scope and the final document readers download.
All research, figures, approvals and events are invented. NIST and Adobe references provide technical context rather than a legal judgement about a real report or a claim of malicious alteration.
Chapter index
- 1. The approval badge belongs to yesterday’s report
- 2. What a signature can establish, and what it cannot
- 3. Two versions can both contain valuable work
- 4. NIST explains integrity and claimed signer identity
- 5. Adobe explicitly distinguishes signed and current versions
- 6. A matching name is not a matching file
- 7. Do not make an ordinary correction look like malicious tampering
- 8. The response protects the reader’s next decision
- 9. Build the signature check into publication, not just archiving
- 10. Practice: an authentic signature, the wrong version
1. The approval badge belongs to yesterday’s report
A fictional education research institute finishes a study about students using a digital revision programme. Its research lead approves version 4 of a report and signs that PDF through the institute’s authorised digital-signature workflow. The signed object is preserved with a date and corresponding verification record in its controlled repository.
Before online publication, an editor corrects a chart’s axis unit and adds a new appendix explaining a limitation. The revised file becomes version 5. These may be genuine improvements, but the new content has not yet passed the documented research sign-off. The publication portal nevertheless displays a green ‘approved and digitally signed’ badge copied from version 4.
The original signature may be genuine for the original document. The problem is the unsupported assertion that its approval extends unchanged to the newer public file. The Gold Standard connects a signatory’s authority, the precise data signed, later revisions and the object a reader actually downloads.
2. What a signature can establish, and what it cannot
File R1 is the original approved version 4. R2 contains its cryptographic signature and relevant verification evidence. Change record R3 describes the editor’s corrections; R4 is the version 5 public PDF. Web record R5 displays the inherited approval badge. These are different records with different significance.
Digital signatures can authenticate the claimed signatory and help detect alteration of covered data under applicable cryptographic verification. They do not automatically approve every later item with the same filename, title or visible graphic signature. A file may carry a signature image without providing genuine cryptographic evidence at all.
Some PDF signing and certification arrangements permit defined kinds of later changes and can retain validity under specific conditions; others may report changes or problems. Therefore management should not assert that every post-sign edit necessarily invalidates every possible signature. It must identify the signed revision, the actual permitted change rules and the status of the current file.
GS-R02 defines the scope of the authorised approval and GS-R08 asks whether the public claim matches that exact scope.
3. Two versions can both contain valuable work
Version 4 may be an authentic, appropriately reviewed research product. Version 5 may correct an honestly mistaken figure axis and disclose an additional caveat readers should know. There is no need to presume the editor intended to deceive merely because the later version lacks the expected final authorisation.
The substantive question is whether each changed part affects an interpretation the research lead or institution is responsible for approving. A one-word change in units can matter more than several pages of formatting. A new limitation may make the report better science while still needing a recorded decision before the organisation labels it signed.
The centre should preserve both versions and compare the actual changes rather than overwrite the earlier file to conceal a discrepancy. A valid original signature does not justify hiding useful corrections, and useful corrections do not justify attaching an approval badge whose basis belongs to different data.
The Gold Standard seeks trustworthy revision history, not inflexible immutability of every published document.
4. NIST explains integrity and claimed signer identity
The US National Institute of Standards and Technology’s Digital Signature Standard, FIPS 186-5 specifies algorithms for signatures that help detect unauthorised modifications to signed data and authenticate the person or entity represented by the signing credentials.
NIST’s FIPS 204 Module-Lattice-Based Digital Signature Standard describes another signature algorithm family and the same broad goals of data integrity and authentication. This fictional university case does not claim its PDF uses FIPS 204 or that meeting a cryptographic algorithm specification establishes every institutional approval requirement.
The sources reinforce that the assurance applies to the data and verification process, not to a logo printed on a page. A green status copied from an older revision must be checked against the actual file. Under GS-R03, the institution also needs an owner authorised to interpret the relevant approval scope and release decision.
5. Adobe explicitly distinguishes signed and current versions
Adobe Acrobat’s Compare Signed Document Versions guidance explains that users can examine the signed revision and compare it with the current PDF to identify changes made after signing. This reflects a practical difference between a signed historical version and the document currently open for reading.
Such comparison can identify modifications but does not automatically determine their scientific appropriateness, legal effect or acceptability under the particular institution’s approval policy. A change may be allowed under some technical circumstances yet still require academic review for public release.
The institute should not use an archived signature verification result as a universal status for every subsequently uploaded copy. Readers need to know what was signed and what changed. An accepted scientific revision and a technically valid signature are distinct properties, and an accurate portal can explain both without misrepresenting either.
6. A matching name is not a matching file
Two documents can both be named final-report.pdf while differing in an axis label, appendix, embedded data or signature information. A browser link or public catalogue can easily retain the same title even when its contents change. Filename equality is therefore weak evidence of file identity.
A suitable cryptographic digest or other document fingerprint can help test whether two files contain the same bytes under a specified method. Different fingerprints show the files are not bit-for-bit identical; they do not tell a reviewer which scientific version is best or whether a particular edit was authorised. A matching signature graphic similarly cannot substitute for cryptographic validation.
The Gold Standard links each revision, signature and release decision to a controlled file identity. That record must survive copies between the editor, institutional archive and publishing platform. A portal advertising ‘signed’ needs the actual verified basis for the current download, rather than an old approval stored under the same filename.
7. Do not make an ordinary correction look like malicious tampering
The institute’s editor may have corrected a genuine unit mistake and added a useful research limitation in good faith. Those actions do not automatically establish a security compromise, a research-integrity violation or an intention to mislead. The failure described is the absence of a demonstrated sign-off for version 5 combined with a public badge that implies one.
An authorised reviewer should preserve the signed source, current public file, change log and applicable signing and publication rules. The institution should explain the discrepancy proportionately rather than accusing the editor or asserting that cryptography itself failed.
Conversely, management must not quietly backdate a signature or make the old file appear to have included a new appendix. That would corrupt the record that permits readers to evaluate what was known and approved at different times. GS-R09 requires a truthful incident status; GS-R10 requires correcting the publication claim and the workflow that created it.
8. The response protects the reader’s next decision
The responsible publication owner pauses or qualifies the statement that the current PDF is wholly covered by the former signature. The research lead assesses the axis correction and appendix under the institute’s accepted research-review procedure. A separate technical officer can verify the signature and file differences without deciding the scientific merits.
A public clarification can say version 4 was signed, version 5 contains later edits and the approval status of that newer revision is being reviewed. It should not state that the entire current report was digitally signed unless the correct evidence supports that assertion. Any needed correction notice or version replacement follows legitimate institutional policy.
Researchers and students may already have cited version 5. Keeping a traceable revision history lets them identify what was downloaded and whether a particular claim changed. A clear status improves scholarship without unnecessarily hiding information whose scientific significance can be explained.
9. Build the signature check into publication, not just archiving
The institute revises its workflow so scientific approval, editorial preparation, technical signature validation and final publication refer to the appropriate same file version where that is the required guarantee. Different roles may own each stage, but the interface must keep the meaning of approval intact when the file changes.
An editor should be able to make legitimate corrections without accidentally inheriting an approval that no longer applies. The system may use a current file identifier, verified signature properties, attributable revision record and an explicit decision for accepted post-sign changes. Its choice must reflect the real signing method and institutional rules, not a generic ‘all changes forbidden’ shortcut.
One acceptance exercise signs a fictional version 4, creates an unsigned version 5 and attempts to publish version 5 with the old badge. The check must catch the mismatch. A second exercise contains a legitimately permitted change under a defined approved method so the system can accurately describe its status without creating a false alarm.
GS-R07 protects the handoff from researcher to publishing team. GS-R12 ensures a new PDF tool or upload process does not silently detach the signature evidence from the public file.
10. Practice: an authentic signature, the wrong version
Exercise. A research lead digitally signs version 4 of a report. A later version 5 corrects the axis units in one chart and adds an appendix. The website serves version 5 while displaying ‘entire report signed’ from version 4’s verification record. Identify the supported fact, unsupported claim and required release control.
A reasoned finding is: ‘There is evidence of a genuine signature on version 4. The present record does not establish that the current version 5 content falls within that signature’s covered data, permitted modifications or the institution’s accepted scientific approval. The website must distinguish the signed revision from later changes until the appropriate review is complete.’
The next test must compare the signed version, applicable post-sign permissions and final downloadable object rather than relying on its filename or inherited badge. The Gold Standard lesson is that an authentic seal carries authority over particular material, not every future document that resembles it.
Technical references and further Gold Standard casework
- NIST · Digital Signature Standard FIPS 186-5
- NIST · Module-Lattice-Based Digital Signature Standard FIPS 204
- Adobe Acrobat · Compare Signed and Current Versions
- Gold Standard · Information and Record Integrity
- Gold Standard · Decision Records and Exception Registers
- Gold Standard · Core Requirements and Evidence
- Case 08 · Changed Procedure after Process Testing
- Case 36 · Research Software Version Reproducibility
- Case 41 · Archive Files Copied but Retrieval Failed
- Gold Standard Casebook and Decision Practice
Authenticity, permitted post-sign modifications and institutional release authority must be evaluated under the actual signature system and governing policies. GS-R01–GS-R12 is eduKate’s educational management framework.