A corpus becomes infrastructural when its multiplication of objects is accompanied by an equivalent multiplication of relations, addresses and operations. Ten thousand isolated documents remain ten thousand isolated documents; ten thousand persistently identified entities connected through typed metadata constitute a different epistemic object. The decisive threshold is therefore relational rather than numerical. Scale begins to matter when accumulation can be traversed, queried, recombined and verified. Identification supplies the first condition of this transformation. Juty et al. show that unique, persistent and resolvable identifiers establish addressability: an object can remain the same referent even when its location changes. Cousijn et al. extend this logic by demonstrating that persistence becomes substantially more valuable once PIDs are connected through metadata. The resulting graph converts naming into topology. A paper, dataset, person, institution or instrument no longer exists merely as a record; it occupies a position in a relational field whose connections can themselves become evidence.

Relational scale also changes what publication means. Haris, Stocker and Auer expose the limits of a scholarly system in which the PDF remains the dominant terminal object. Machine-actionable knowledge requires claims, contributions, versions and provenance to acquire structured representation. DOI-based identification of ORKG Papers pushes citability below the level of the conventional document while preserving scholarly accountability. The archive becomes granular without becoming unstable: versions can proliferate because provenance preserves their sequence. Machine actionability intensifies this shift. Islam distinguishes readability from interpretability and interpretability from actionability. A machine may parse structured content without understanding the type of entity it encounters; it may identify that type without knowing what operations are permitted. PID-level metadata therefore acts as a thin operational membrane around heterogeneous digital objects. Scale can then be processed without requiring every object to possess the same internal form. Interoperability is achieved not through homogenisation but through a shared layer of addressable distinctions. 

OpenAlex demonstrates the consequences at planetary scholarly scale. Hundreds of millions of works become tractable because they are modelled as nodes connected to authors, institutions, venues and concepts, with canonical external identifiers binding the graph to other infrastructures. Numerical expansion is subordinated to schema. This is crucial: the meaningful metric of a knowledge corpus is not simply how many elements it contains, but how many of those elements can be distinguished, linked, resolved, cited, versioned and computationally traversed. Density without legibility produces opacity; legibility without relation produces a catalogue; relation plus persistent addressability produces infrastructure. Socioplastics can therefore understand its expanding corpus through relational scale rather than celebratory counting. Nodes, papers, PDFs, DOIs, indexes, slugs, references and public interfaces become consequential insofar as they form a navigable epistemic topology. An 11,000-node threshold matters not as proof of importance but as evidence of a system whose internal differentiation now requires explicit governance: stable identifiers, metadata layers, provenance, machine-readable relations and durable public access. Corpus growth becomes architectural when every additional element can reinforce the intelligibility of the whole. Relational scale adds a precise criterion to Socioplastics: accumulation becomes field only when quantity is converted into addressable, interoperable and operational structure.