Interoperability is usually presented as an unquestioned good. Systems should communicate, repositories should exchange records and vocabularies should align. Yet connection often has a hidden cost: different things are made artificially equivalent because exact equivalence is technically convenient. A local concept is mapped onto a generalized category. A translation suppresses an asymmetry. A rich metadata role becomes a generic field. The resulting systems exchange information efficiently while losing part of what that information originally meant. A stronger form of interoperability would treat difference as information rather than noise. Two concepts may overlap without being identical. They may perform similar functions while belonging to incompatible theoretical traditions. One may be broader than another. Translation may work better in one direction than the other. A correspondence may be valid for one task and misleading for another. Recording these relations introduces complexity, but it also produces more faithful knowledge. Technical interoperability remains indispensable. Shared identifiers, compatible formats, APIs and metadata crosswalks allow distributed research to function. Yet syntactic exchange does not guarantee semantic understanding. Two systems can exchange a field named “subject” while meaning substantially different things by it. A DOI can establish that two records refer to the same publication without proving that every description of that publication is equivalent. Identification creates continuity of reference; interpretation still requires relations and context. The same problem becomes more consequential in retrieval and generative systems. When heterogeneous sources are grouped primarily through lexical or embedding similarity, related concepts can collapse into a statistical average. Historical versions, local formulations and minority vocabularies disappear because normalization rewards the dominant semantic neighbourhood. An interoperability layer that records partial equivalence, directional translation, disagreement and uncertainty could preserve this plurality instead of smoothing it away. The objective is therefore not one universal vocabulary capable of replacing every local language. It is a federation in which several languages, disciplines and communities can exchange knowledge while keeping visible the structures that make their concepts meaningful. Interoperability reaches maturity when a system can represent not merely that two things connect, but how, for what purpose and with what loss. Communication across difference is stronger than communication achieved by deleting difference.
Berners-Lee, T. (2006) ‘Linked Data: Design Issues’, W3C.
Euzenat, J. and Shvaiko, P. (2013) Ontology Matching. Springer.
Gruber, T. R. (1993) ‘A Translation Approach to Portable Ontology Specifications’, Knowledge Acquisition, 5(2), 199–220.
Star, S. L. and Griesemer, J. R. (1989) ‘Institutional Ecology, Translations and Boundary Objects’, Social Studies of Science, 19(3), 387–420.
World Wide Web Consortium (2009) SKOS Simple Knowledge Organization System Reference.