Exact-byte continuity
Given the same algorithm and the expected digest, a matching digest provides strong evidence that the compared byte sequence is unchanged.
Integrity is a boundary, not a verdict. Preserve the exact artifact—and refuse to pretend exactness proves what the artifact says.
What SHA-256 and preservation fixity can establish, what they cannot establish, and why provenance must remain separate from truth.
A HASH CAN TELL YOU THE FILE CHANGED. IT CANNOT TELL YOU THE FILE WAS TRUE.
Sixty-four hexadecimal characters can look like a verdict. They are not. SHA-256 is extraordinarily useful because it lets us compare exact digital content with a compact digest. That is a claim about bytes. The mistake begins when integrity is promoted into authorship, accuracy, morality, provenance, review, or truth.
This article describes cryptographic hash and digital-preservation practice at a high level. It does not claim that hashes alone authenticate authorship, establish legal admissibility, or prove that preserved content is factually correct.
NIST's Secure Hash Standard specifies algorithms used to generate digests of messages. The digest gives you a compact value that can later be compared with another computation over the same bytes.
The Library of Congress uses the same idea in preservation practice: fixity metadata and cryptographic hash values are recorded and compared to monitor whether stored digital content has changed. PREMIS preservation guidance describes the same pattern—calculate a digest earlier, calculate it later, compare the two.
That is powerful. It is also bounded.
Given the same algorithm and the expected digest, a matching digest provides strong evidence that the compared byte sequence is unchanged.
A digest can help detect corruption or unintended modification during copying, storage, packaging, or distribution.
Two files with different digests are not byte-for-byte identical under that algorithm.
Recorded digests make later fixity checks auditable when the manifest itself has trustworthy provenance.
A false report can be hashed perfectly.
A digest does not tell you who created the file.
The first file you received may already be altered, incomplete, or misattributed.
A preserved screenshot can still omit what happened before or after it.
A checksum does not mean a qualified human evaluated the content.
Exact bytes do not turn a document into a controlling source merely because the copy is stable.
For consequential archival work, keep several questions separate:
Filename, byte size, media type, digest.
Source URL, custodian, transfer route, retrieval date.
Title, author, date, edition, scope.
Reviewer, method, competing evidence, limitations.
Correction, supersession, withdrawal, replacement digest.
Missing custody, uncertain authorship, absent context, unresolved contradiction.
Hash the artifact. Preserve the source. Name the authority. Record the review. Then keep those claims separate.
Hashes can participate in much stronger systems. Signed manifests, trusted timestamping, controlled repositories, content-addressed storage, and independently witnessed provenance can combine byte integrity with evidence about custody and identity. The point is not that hashes are weak. The point is that each layer proves a different thing.
Cryptographic standards also evolve. NIST has announced plans to revise FIPS 180-4. Long-lived archives therefore need algorithm policy and migration planning rather than treating one digest format as immortal.
No digest can reconstruct missing context by itself. When custody before the first known hash is uncertain, say so. When the authority changed the file and published a new digest, preserve both states rather than pretending one timeless artifact existed.
Initial publication. Review when NIST replaces or materially revises FIPS 180-4. Submit corrections through Corrections.
PRESERVE THE BYTES. DO NOT WORSHIP THE DIGEST.
A hash is a witness to exactness. It is not a witness to meaning.