The first location is the canonical sealed file — the Core TXT, the DOI-registered object, the dataset entry. Here the access block is frozen at the moment of sealing. It records the corpus's infrastructure as it existed when that layer was closed. This version is historical and citational: it tells a future reader what the system looked like at a specific threshold. It should never be updated. Its value is precisely its fixity. The second location is a living public interface — a dedicated index page, a pinned post, a repository README, a dataset card. Here the access document is not sealed but maintained. It reflects the corpus as it currently stands: the most recent DOI objects, the active platforms, the current node count, the latest Core layers, the active dataset version. This version is navigational and operational. It changes when the corpus changes, and its reliability depends on the discipline with which it is maintained. The third location is the corpus's own index structure — the MasterIndex, the CenturyPack headers, the Layer maps. Here the access document is distributed, embedded in the architecture itself rather than appended to it. Every CenturyPack that opens with its position in the scalar hierarchy and closes with its related layers and persistent identifiers is performing a local access function. The corpus becomes self-describing at every scale. These three locations are not redundant. They serve different temporal functions: the sealed file records the past, the living interface describes the present, and the embedded index enables navigation at any moment. A corpus that maintains all three has solved the access problem architecturally.
