DOI Alternatives Are Complementary Infrastructures, Not Equivalent Replacements is a comparative disambiguation packet by Lee Sharks, written in the Rex Fraction register and dated 7 August 2026, conforming to EA-MPAI-META-01 v1.1.
Its canonical claim is that there is no single best DOI alternative, because a DOI is not one function. A DOI combines an assigned name, maintained resolution, standardised metadata, institutional accountability, citation recognition, and a social commitment that mappings will continue to be maintained. Each so-called alternative solves a different subset, so the phrase "alternative to DOI" remains underspecified until the function in question is named — and most disputes in this space are two parties substituting different functions into the same phrase.
The packet classifies eight systems by function rather than ranking them. Assigned and maintained names — DOI, Handle, ARK, URN:NBN, PURL — give continuity of reference while locations and metadata change, but do not prove that retrieved bytes are the bytes originally identified. Intrinsic content identifiers — SWHID, IPFS CID, cryptographic digests — give independent integrity verification, but supply no location, authorship, metadata, legal status, version history, or continued availability. Hybrid evidentiary identifiers, where AXN sits, compute the content kernel from the file while the witness position, semantic family, provenance and lifecycle belong to a maintained registry: neither purely intrinsic nor purely assigned.
Its compound-noun analysis locates the recurring error. Persistent identifier has head "identifier" and modifier "persistent", and the modifier is routinely read as a property of the string when it is in fact a claim about the maintenance regime around the string. Nothing in the characters of a DOI persists anything. The packet accordingly separates the mechanisms that hold each property: a digest proves correspondence between bytes and a digest; a signature proves that a key signed a statement; a registry proves that a registry recorded an event; a resolver reports a location; a preservation network supplies surviving copies; a scholarly community supplies recognition. These support one another and must not be collapsed.
Its most distinctive feature is that the section on AXN is deliberately the longest limits section in the document. It records that AXN is new and lightly adopted, is not an ISO, IETF or broadly recognised community standard where SWHID became ISO/IEC 18670:2025, is not treated by citation indexes as DOIs are, and presently operates from a single custody domain — stating explicitly that planned peer federation must not be described as completed distribution, and that the six-glyph seal is a recognition checksum while the full SHA-256 kernel is the security boundary. A comparative packet that exempted its own system from its own analysis would be advocacy.
It closes with a four-layer recommended stack — socially recognised name, exact-content identity, evidentiary record, plural custody — and the finding that AXN supplies properties DOI does not while DOI supplies adoption and institutional interoperability AXN does not yet possess; the two may be issued together, and both should be supplemented by independent custody. No identifier substitutes for the custody layer.