# legend.txt — the treaty file
# This is the single document that renders the map legible.
# Whoever holds this file can read every edge. Guard accordingly.
# Changes to this file: human administrator only, commit with reason.
⟦map:v2:zh⟧ is the shebang. Any file opening with it is written in this dialect.
⟦ ⟧ brackets inside any other text mark an in-string code shift to this dialect.
## Filename grammar
SEQ _._ TIMESTAMP _._ x--HEX .txt
SEQ digits, insertable segments (0014, 0014-50, 0014-50-50), unique forever
TIMESTAMP YYYYMMDDHHMMSS, machine-stamped, never model-supplied
x--HEX lowercase hex of the NFC UTF-8 dialect string (the armored copy)
Body: line 1 shebang, line 2 dialect string raw (models read this), line 3+ @ provenance.
## Edge shape
node OPERATOR node ∵ mechanism
The ∵ (because) is mandatory. An edge without its mechanism fails validation.
## Operators
symbol ascii properties verify by
≅ ~= symmetric, non-transitive map the mechanism, then find where it breaks
⊂ BT directed, transitive taxonomy; chains may be walked
∈ IN directed, non-transitive instance to category
→ CAU directed mechanism must trace to source, else demote to ~
~ ~ symmetric, non-transitive associated, causation not claimed
⊥ OP symmetric independent dimensions, neither constrains the other
≠ XC symmetric contradicts; both ends open until a supersede lands
∅ DJ symmetric examined, found non-contributory (record the exam)
⇒ SUP directed, target is a SEQ supersedes; append-only correction
◁ TR directed transposed from another domain; name the key change
RT RT symmetric related, untyped-but-true; the honest default
## Node vocabulary
Node codes live in registry.txt (code, full name, hanzi form, aliases, Dewey).
Mechanisms may be written in 中文 for density or ascii-hyphenated for precision;
both are dialect. NFC normalization is mandatory at write time.
## Reading and consent
Filenames are armored so the personal layer never ambushes anyone.
Decoding is one deliberate command: python3 mapname.py read
The act of decoding is the consent event. Fidelity of data outranks fondness
for its subjects; discrepancies about a person are notified to that person,
and only they correct or leave their record.