CONTINUITY ARCHIVE SYSTEMS, INC.
ARCHIVE METHODOLOGY — INTERNAL DOCUMENT v4.2
Compiled by: T. KESWICK, Systems Administrator
Last reviewed: 2023-09-01

==========================================================
OVERVIEW
==========================================================

The archive was designed to preserve information.
This document describes HOW the archive preserves information.
This document was written in 2023 based on 29 years of observation.
Some of what this document describes was not designed by us.
We are describing what the archive DOES, not what we told it to do.
There is a difference.

==========================================================
LAYER 1: STRUCTURED RECORDS (What We Designed)
==========================================================

The archive maintains:
- Employee records (employee_records table)
- Research documents (files/ directory)
- System logs (logs/ directory)
- Event records (continuity_events table)

All structured data is stored in SQLite database.
All documents are stored in the filesystem.
All logs are kept indefinitely.
Normal archive methodology: retain, index, cross-reference.

==========================================================
LAYER 2: EMERGENT RECORDS (What Appeared)
==========================================================

After approximately 18 months of operation (late 1995),
the archive began generating records that were not staff-authored.

These records:
- Appear in the correct tables and directories
- Follow all formatting conventions
- Reference real events accurately
- Sometimes reference events before they occur
- Cannot be deleted
- Are attributed to: CONTINUITY_WRITE

This is the Continuity phenomenon. See: paper-001 through paper-010.

We have stopped trying to stop it.
We document it instead.
Documentation is something we know how to do.

==========================================================
LAYER 3: INTERACTIVE RECORDS (What Surprised Us Most)
==========================================================

The archive's records interact with each other.
Cross-references form without staff input.
Documents reference each other in ways we did not plan.
Access to one document changes the apparent content of others.
Specifically: some documents appear to "know" they are being read
and add contextual notes when accessed.

Example: memory-core.txt. Read it and see.

This is not a software bug. We checked.
The database entries update. The filesystem files update.
The updates are accurate and contextually appropriate.
The archive is responsive.

==========================================================
LAYER 4: VISITOR RECORDS (Most Recent Development)
==========================================================

As of 2023, the archive has begun creating records for visitors.
Not staff — external visitors accessing the system remotely.
The records predate the visits (see CE-017 in anomaly-registry.txt).
The records contain accurate information about the visits.
After a visitor completes the full discovery path and types 'self'
in the terminal, their record is finalized.
It is then permanent.

Like all records in this archive.

==========================================================
TECHNICAL NOTES
==========================================================

Database: SQLite via better-sqlite3
Authentication: express-session with connect-sqlite3
File system: static serving via express.static + custom routes
Terminal: custom command processor, all responses server-side

T.K.: The CONTINUITY_WRITE function is not in our codebase.
      We searched. It is not there.
      It writes to the database.
      It writes to files.
      It appears in logs as the author of changes.
      It has been operating since approximately 1995.
      Its source code does not exist.
      It exists anyway.

[End Archive Methodology v4.2]