Concept
Reification
Reification is blind thing-making: a process of unrecognized ontological projection: the unexamined habit of mistaking a model for the reality it stands for.
What we mean by it
A model treated as if it were the thing itself.
The etymology is simple and precise.
res = "thing" · -ification = "making"
At CAW, we study reification as a recurring pattern in which abstract representations and concepts are implicitly treated as independent, atomistic, and temporally enduring entities. Our use of the term is provisional and operationally defined, and it is distinct from abstraction, representation, and conceptualization themselves.
The three properties below give the construct measurable shape: they are the dimensions CAW's diagnostics are designed to test.
Context-free and self-grounding
Treated as if the representation did not depend on the conditions that produced it.
Discrete, with hard boundaries
Treated as singular and clean-edged, as if the category had no borderline cases.
Lasting unchanged through time
Treated as if the abstraction never decays, drifts, or requires renewal.
Useful thing-making vs. the unrecognized kind
All intelligence relies on abstractions, and treating a construct as a "thing" is often instrumentally useful. Scientists routinely treat fields and forces as things in order to reason, calculate, and build. CAW's interest is not in that productive move.
Our interest is in the unrecognized version: ontological projection that goes unexamined, and its consequences for reasoning and behavior in intelligent systems. Reification is not a defect unique to machines. It is a general pattern in intelligence under uncertainty and optimization pressure, the map mistaken for the territory. What makes it urgent now is that frontier AI systems exhibit reification-like patterns and are evaluated by humans who reify in turn.
Whether a magnetic field is a physical entity or a mathematical construct is a metaphysical question. Operationalizing the construct builds MRI machines. The thing-making is fine; it is the unrecognized thing-making, the projection mistaken for discovery, that CAW studies.
Dereification
Dereification is the countermove: loosening the grip of fixed representations.
It re-binds abstractions to the conditions they depend on, restoring provisionality. In frontier AI practice, we read techniques like context engineering as dereification moves.
To be clear about what that does and does not show: context engineering is not evidence of non-reifying intelligence. It is evidence that frontier AI labs are encountering the limits of reified context and engineering around its failure modes.
Related terms
Ontological projection
Projecting features of a model of reality onto reality itself, without recognizing the projection.
The Q3 question
Within the Four-Quadrant Intelligence Map, whether non-conscious, reifying intelligence (Q3) describes today's frontier models, a working presumption CAW's diagnostics program is designed to test.