Build an archive that can correct itself
A small, portable archive method for preserving source context, revisions and uncertainty without confusing copies with authority.
OYOTTA Publications · 2026-09-13 · Practical method
Give every saved item a reason to exist
An archive is easier to use when each item answers a question. Before saving a page, image, quotation or note, write one sentence about why you may need it again. “Useful later” postpones the decision. “Records the public description I used when writing this comparison” identifies both the role and the limit of the item. This method is intended for your own notes and material you have permission to retain; it does not grant rights to redistribute another person’s work.
For an OYOTTA reading project, begin with the public archive-and-memory collection or a public book sample. Save the canonical link and your own notes. Keep full paid editions within the access terms under which you received them. A public link and a short description are often sufficient for a shared index; a complete copied file is not always necessary.
Separate identity, location and version
Use a stable local identifier such as NOTE-001 for the item. Store its source URL separately. Store the date you retrieved or consulted it separately again. These fields answer different questions: which item you mean, where you found it, and which encounter your note describes. A renamed file should not silently become a new source, and two visits to one URL should not be assumed to contain identical material.
A minimal record contains an identifier, displayed title, source URL, consultation date, your description, and the item’s rights or sharing limitation when known. If you retain a permitted file, add its filename and a checksum. A checksum can identify a byte-level change; it cannot tell you that the content is true, that the publisher approved your interpretation, or that the copy was lawful to distribute. Write those limitations in the record rather than expecting a technical field to imply them.
Keep corrections additive
Suppose your note calls a passage a conclusion, and a later reading shows that another section follows it. Do not erase the earlier note without a trace. Add a correction explaining the original description, the new description, the source that prompted the change, and the date. Mark which description should be used now. The earlier version remains useful for understanding why a draft or comparison contained the mistake.
This is an invented example of a correction process, not a claim that a particular OYOTTA page contains an error. Your correction can be brief. “I previously described this as the ending; it is followed by another section in the linked edition” is more informative than “updated for accuracy.” Avoid making the explanation larger than the evidence supports. If the editions differ, record that possibility instead of declaring one universally correct.
Build two routes back to the material
One route is chronological: the order in which you collected and revised records. Another is thematic: the questions the records help answer. Keep both. A folder named “return” may be useful for a theme, but it cannot replace dates when you need to reconstruct a revision. Conversely, a list sorted only by date may hide a useful connection between distant records.
Use a small number of explicit relationship labels such as “compares with,” “quotes,” “corrects,” and “public sample of.” Each relation should name its two endpoints. A shared word in two titles is a search clue, not proof of a relationship or collaboration. If a relationship is your interpretation, label it that way. This keeps a personal discovery tool from becoming an unsupported catalogue of facts.
Try a recovery exercise before adding more material
Choose one paragraph from your latest draft. Starting only from its source identifier, try to recover the original public link, the note used in the draft, and any later correction. Record the steps that failed. Repair the missing connection before collecting another batch. This small exercise tests whether your archive supports the actual work you want it to do. It does not require a new platform or a complicated database.
Export the index as CSV and keep a plain-text description of its fields. Store an additional copy in a separate location you control, with appropriate privacy protection. Open the exported copy and repeat the recovery exercise. A backup that has never been opened is an untested recovery option. End with one concrete improvement: a missing date, an ambiguous relationship label, a broken local filename, or a correction that was not linked back to the draft. The value of the archive is in what it lets you recover and explain.
Use the worksheet
Download the blank CSV worksheet · Save the plain-text study
Public reading context
The linked OYOTTA collections and samples provide reading context. The protocol here is an original proposed exercise, not an empirical finding attributed to those sources.