Ontological commitment

An ontological commitment is the set of entities, relations, and properties that a system — whether a database, a scientific theory, or a language — assumes to exist in order to function. It is not a claim about what is real in the universe; it is a claim about what the system must treat as real in order to make sense of its own operations. When a database schema declares that 'Customers' exist and have 'Credit Scores,' it is making an ontological commitment that shapes every query, every report, and every business decision derived from that data.

The term originates in analytic philosophy, where Willard Van Orman Quine argued that to be is to be the value of a bound variable — in other words, what a theory quantifies over is what it is committed to. In systems thinking, this philosophical precision matters: a system's ontological commitments are the hidden axioms that determine its behavior. Change the commitments and you change the system, even if every line of code remains the same. The most dangerous systems failures are not code errors but ontological mismatches — moments when the world behaves in ways the system's implicit ontology cannot represent.

Ontological commitments are rarely explicit. They are embedded in schemas, data models, user interfaces, and organizational processes. Making them visible is a form of systems archaeology: the excavation of assumptions that have been buried so deeply that they are mistaken for facts. The discipline of ontological engineering — still more aspiration than practice — aims to design systems whose commitments are explicit, revisable, and accountable.

The failure to recognize ontological commitment as a design choice is the original sin of software engineering. Every system encodes a theory of reality, and most encode bad theories without knowing it. A system that assumes customers are individuals with fixed identities will fail in markets where identity is fluid and collective. A system that assumes causation is linear will fail in ecosystems where causation is recursive. The ontology is not a layer you can add later; it is the foundation, and foundations are the hardest things to change.