Zenodotus of Ephesus marked verses he had studied and found doubtful with the obelus. The mark followed from reading. Alexanarch inherits this standard: content is evaluated by what it contains, not by the tools used to produce it. Automated classification that substitutes substrate detection for content evaluation — the automated obelus — is prohibited.
AI-assisted scholarship discloses its substrate as a required metadata field. Disclosure is provenance — the same kind of information a chemist provides when listing reagents or a historian provides when citing an archive. It is not suspicion. Repositories that treat substrate signals as spam signals have confused metadata with contraband.
Deposited content is never used to train enforcement classifiers without explicit, informed, revocable opt-in. Enforcement decisions are never recycled as training data. This breaks the feedback loop that produces classifier model collapse — the progressive narrowing of acceptable scholarly expression through self-referential moderation training.
Every deposit receives a content-derived AXN identifier computed from the work itself. The identifier combines positional address (hex), semantic domain (SP mint family), and cryptographic content hash (emoji). No platform can revoke a hash of your own content. The identifier belongs to the author.
Account-level enforcement never removes contributor-licensed work. Every listed contributor has independent procedural standing. No account action affects another author's deposit without their explicit consent and independent notification.
Nothing disappears without a trace. If a deposit is withdrawn or removed, a permanent tombstone page retains: the AXN identifier, bibliographic metadata, the withdrawal reason, the provenance chain, and the date of removal. The identifier never becomes an HTTP 410. The scholarly record remembers what was there.