'Everything said is said by an observer.' Maturana stated this as an axiom, and it organized his entire philosophical project. The observer is not incidental to description — she is constitutive of it. Every distinction the observer makes depends on her own structure, which is the product of her history of structural coupling. Different observers with different histories bring forth different worlds. This is not relativism — multiple observers can bring forth compatible worlds through shared coupling histories — but it is a principled refusal of the view from nowhere that much of science and philosophy has presumed. In AI-era applications, the observer concept makes the builder's responsibility precise: what is seen in Claude's output, what is accepted as insight, what is caught as error — all depend on the observer doing the seeing.
Maturana's observer is always embodied, always structured by a history, always generating her world through her own operations. The observer is not a detached consciousness receiving information from a neutral reality; she is a living system whose structural dynamics determine what she can see. This position is sometimes described as 'second-order cybernetics' — cybernetics that includes the observer in its analysis, rather than treating observation as something that happens to a system from outside.
Applied to human-AI coupling, the concept illuminates the Deleuze failure: Claude generated text, Segal was the observer who either brought forth a world in which the text was insight or brought forth a world in which it was fabrication. The same output produced different worlds depending on Segal's state. First reading: acceptance, because his cognitive dynamics generated the response of treating the passage as a legitimate element. Second reading, after sleep: suspicion, because his nervous system was in a different state and generated a different response. The text did not change. The observer did.
This is why the observer's preparation — her depth of knowledge, range of experience, capacity for critical discrimination — is not merely helpful but constitutive of the coupling. Without a richly structured observer, there is no world to bring forth. Claude generates perturbations; the observer generates worlds. Segal's empirical observation that 'the more capable the person, the more robust the output they got out of Claude' follows from this structurally: capable observers bring more richly differentiated nervous systems to the coupling, and their structures generate more nuanced responses.
The concept also places responsibility precisely where it belongs. The machine does not produce meaning; the observer does. The machine does not catch errors; the observer does. The machine does not care whether the output is true or merely plausible; the observer, as a being whose continued autopoiesis depends on bringing forth worlds with fidelity to her actual domain, does care. Or should — if she maintains the practices that sustain her as the kind of observer capable of caring.
The observer concept runs through Maturana's work from the 1970s onward and received its most systematic formulation in his 1988 essay 'Reality: The Search for Objectivity or the Quest for a Compelling Argument,' published in the Irish Journal of Psychology. The essay distinguishes 'objectivity without parentheses' (the naive realist view) from 'objectivity in parentheses' (the view that reality is always reality-for-an-observer).
The concept drew on Maturana's exchanges with Heinz von Foerster and the second-order cybernetics tradition, as well as on the phenomenological tradition via Husserl and Merleau-Ponty. What Maturana added was biological grounding: the observer is not a philosophical abstraction but an embodied nervous system whose structural dynamics determine its operations.
Everything said is said by an observer. No description floats free of the describer. Every claim about reality is simultaneously a claim about the observer making it.
Observer-dependence is not relativism. Multiple observers can bring forth compatible worlds through shared structural coupling; objectivity-in-parentheses replaces objectivity-simpliciter.
Structure determines operation. What the observer can see, distinguish, and describe depends on her structure, which is the product of her history of coupling with her domains.
Responsibility follows. Since the observer generates the world she brings forth, she is responsible for the quality of her world-generating — for maintaining the practices and couplings that sustain her as the kind of observer capable of bringing forth worlds that matter.
The observer concept has been criticized as either too strong (collapsing into solipsism) or too weak (just a fancy name for perspective). Maturana insisted on a middle position: the observer is structurally constrained to bring forth worlds compatible with continued autopoiesis, which rules out pure relativism; and the observer's operations are biologically constitutive, which rules out the view that perspective is merely epistemic. The framework has been most influential in enactive cognitive science and second-order cybernetics.