Look. Predict. Verify.

1. Make a prediction

Which files will match: original, renamed copy, backup copy, one-character change, CRLF, no final newline, empty and empty copy? Which names or visible text might mislead you?

2. Compare individual files

1. Open DigestBench. Keep “Local file” and “SHA-256.”
2. Choose 01-original.txt, leave Expected checksum empty and select “Calculate & compare.” Record the byte count and full hash. Optionally download a receipt.
3. Paste that full original hash into Expected checksum. Choose 02-renamed-copy.txt and calculate again. Record the result.
4. With the same expected checksum, try files 04, 05 and 06. Record match or mismatch and byte count. A mismatch identifies a difference, not its cause.
5. Try 07-empty.txt. Is a zero-byte file the same as “no result”?

3. Review the collection

1. Open TwinTrail. Choose files and select all eight fixtures together, once. Do not select the ZIP, manifest, worksheet or a prior export.
2. Leave “Confirm every byte” checked. Leave Protected keep folder empty. Select “Find duplicates.” How many groups appear?
3. Select “Keep every copy.” Observe the review-copy and potential-byte counters.
4. For this fictional exercise, leave 01-original.txt and 07-empty.txt marked Keep. Uncheck Keep only for files 02, 03 and 08. Do not change files outside these groups.
5. Export CSV or JSON. Review which files are marked Keep. This is a review list, not a deletion operation.

4. Challenge the checksum

This is a fictional untrusted-source scenario. Nobody is claiming a file’s origin or impersonating a publisher.

In DigestBench, calculate file 04’s hash without an expected value. Then use that result as file 04’s expected checksum and calculate again. It should match. Does that establish that file 04 is the intended original? What if someone supplied both the file and the hash?

5. Exit questions

1. Can differently named files contain identical bytes?
2. Can files with the same byte count contain different bytes?
3. Why can two similar-looking text files produce different hashes?
4. What does a checksum match leave unproved?
5. After exporting the review list, how many files has TwinTrail deleted? Why might you deliberately keep identical copies?