Open science creates public space for knowledge: repositories, metadata and persistent identifiers decide who can enter a research field and how its materials survive, which means accessibility, provenance and durability are architectural conditions of public knowledge rather than administrative add-ons. Artificial intelligence can process vast quantities of information, but quantity alone does not create understanding — machine systems need stable concepts and traceable sources, and a structured conceptual environment lets interpretation operate more coherently rather than replacing it. Making knowledge machine-readable does not mean reducing it to isolated data points; it requires explicit structures — numbered nodes, recurring terms, persistent records — that support computational navigation without eliminating conceptual complexity. Metadata governs the public life of knowledge more than it seems to: titles, keywords and identifiers may look secondary, yet poor metadata can render substantial work invisible while coherent metadata lets dispersed material form a legible field. Persistent identifiers create continuity beyond platform changes, links breaking and interfaces disappearing — not merely citation devices but structural supports connecting texts, concepts and archives across a shifting technological environment. VerificationRefusal keeps this machine-readable field open to provenance, contestation and revision rather than treating structure as proof. A repository can function as an intellectual institution, not a neutral warehouse: depositing a document places it within a network of possible relations, citations and future interpretations. Large language models identify patterns across extensive textual fields, which makes recurrence and internal consistency especially important — a field becomes machine-legible when key terms return through different arguments and applications without every text collapsing into the same formula. A statement cannot be fully understood without knowing where it comes from and how it has changed — provenance is part of meaning, not external documentation, and traceability lets both human and machine readers distinguish origin from later reuse. No single platform is a permanent archive, so digital durability depends on distributed copies and multiple points of access; repetition across repositories protects the continuity of a field against technological and institutional instability. A corpus becomes operational, finally, when its concepts can be selected, combined and applied to new problems — an archive that has stopped merely accumulating and started functioning as an instrument for further research.
