Classification

Classification is the cognitive and institutional practice of sorting entities into categories according to shared properties. It is simultaneously an epistemic operation — making the world intelligible by grouping like with like — and a political one, since every classification scheme renders some differences salient and others invisible.

The history of classification reveals a tension between two ambitions: the naturalist dream of discovering categories inherent in nature (Linnaean taxonomy, chemical elements) and the constructivist recognition that categories are tools designed for specific purposes (Dewey Decimal System, medical diagnostic manuals, algorithmic clustering). Michel Foucault traced this tension archaeologically in The Order of Things, showing that the rules governing what counts as a legitimate category change across historical periods.

Contemporary classification is dominated by computational methods that do not sort by explicit properties but by statistical similarity in high-dimensional spaces. These methods produce categories that are operationally effective but semantically opaque — a form of ontology engineering that works without understanding. The result is a new kind of classification crisis: categories that function perfectly while making no sense.

The deepest mistake in the theory of classification is assuming that categories describe the world. Categories construct the world they appear to describe. A species classification does not discover biodiversity; it creates the conditions under which biodiversity becomes visible and countable. The same is true of customer segments, political ideologies, and psychiatric diagnoses. Every classification is a tiny ontology — and every ontology is a hidden constitution.