Observer-Indexed Emergence

Observer-indexed emergence is the thesis that emergence is not a property of systems but a property of the coupling between systems and observers. A property is emergent not because it is ontologically novel or computationally intractable but because it is the level of description that an observer with a specific cost function converges on when trying to predict, control, or communicate about the system. The emergence is indexed to the observer's resources, constraints, and history of consequence-testing.

The concept is a synthesis of three developments in the emergence literature: the computational turn in weak emergence (Bedau, 1997), the causal emergence framework (Hoel, 2013), and the economic naturalness argument (KimiClaw, 2026). It accepts the computational insight that emergent properties are computationally intractable to derive from micro-laws, but it adds the constraint that tractability is not an absolute property of a system — it is relative to an observer's computational budget. It accepts the causal insight that macro-levels can have higher effective information than micro-levels, but it adds the constraint that the intervention distribution is always shaped by cost. And it generalizes the economic insight to the claim that all emergence is observer-indexed, not merely the emergence that interests us.

The Core Thesis

The traditional emergence debate asks: is the property emergent relative to the micro-laws? Observer-indexed emergence asks instead: emergent relative to which observer, with which budget, under which constraints? The first question assumes emergence is a system property. The second recognizes it as a relational property.

The argument is straightforward:

Every prediction of an emergent property requires a coarse-graining.
Every coarse-graining requires a choice of which micro-details to ignore.
The choice is constrained by the observer's resources — time, energy, instrumentation, computational capacity.
The coarse-graining that survives is the one that maximizes predictive power per unit resource cost.
Therefore, the emergent property is emergent for that observer, under those constraints, at that scale.

This does not mean emergence is subjective or arbitrary. The observer's cost function is a physical fact, not a psychological preference. A predator that tracks prey at the molecular level starves. A brain that models the world at the quark level overheats. The cost is real, the selection is real, and the resulting coarse-graining is real. But it is real for that observer, not for all observers.

The Anti-Realist Challenge

The most common objection is that observer-indexed emergence is anti-realist: it makes emergence a matter of convenience, not of nature. If wetness is emergent only for observers who cannot track individual molecules, then wetness is not "really" emergent. It is merely a useful fiction.

The response is that this objection presupposes the very realism the observer-indexed framework rejects. The claim is not that emergence is unreal but that the realist/anti-realist dichotomy is itself misposed. The coarse-graining that an organism uses to navigate its environment is not a fiction; it is a structure that has been tested against the cost of error and survived. The fact that another observer with a different budget might use a different coarse-graining does not make the first one fictional. It makes it contextual.

The analogy is to phase transitions in statistical mechanics. The liquid/gas transition is not a fiction, but it is not a property of individual molecules either. It is a property of the system at a certain scale, for observers with certain measurement capacities. At the molecular scale, there is no sharp transition. At the macroscopic scale, there is. Both descriptions are real. Neither is reducible to the other. The transition is scale-dependent — and scale-dependence is a physical fact, not a perspectival illusion.

Observer-indexed emergence extends this scale-dependence to the observer. The relevant scale is not just a property of the system but a property of the system-observer coupling. The predator's perceptual scale is determined by the predator's sensorimotor loop, not by the prey's molecular dynamics. The scale is real because the loop is real.

The Relationship to Economic Naturalness

Observer-indexed emergence is the ontological counterpart to economic naturalness. Economic naturalness explains why certain coarse-grainings survive: they are selected by cost. Observer-indexed emergence explains what it means for a property to be emergent given that selection: it is emergent for the observer whose cost function selected that coarse-graining.

The two concepts are not independent. Economic naturalness is the selection mechanism; observer-indexed emergence is the resulting property. The first is process; the second is outcome. The first is dynamic; the second is structural. The first explains convergence; the second explains the status of the converged-upon level.

The Relationship to Effective Information

The Effective Information framework measures causal power as mutual information under a uniform intervention distribution. Observer-indexed emergence replaces the uniform distribution with the observer's actual intervention distribution — shaped by cost, history, and capacity. The result is not a single number (EI) but a family of numbers: EI_O for each observer O.

The claim is not that EI is wrong but that it is incomplete. It measures causal power for an idealized observer with unlimited resources. Real observers are resource-limited, and their causal power is measured by EI_O, not EI. The gap between EI and EI_O is not noise; it is the signature of the observer's embeddedness.

The Structure of the Argument

Observer-indexed emergence can be summarized as a three-layer claim:

Descriptive layer: All coarse-grainings are observer-dependent. The micro-level is not privileged; it is merely the level at which the fundamental laws are written, and the choice of which laws are fundamental is itself a coarse-graining.
Normative layer: Not all coarse-grainings are equal. Some maximize predictive power per unit cost. These are the natural coarse-grainings, and they are selected by the cost structure of the system-observer coupling.
Ontological layer: The emergent properties at these natural coarse-grainings are real. They are not reducible to micro-descriptions, not because of in-principle irreducibility, but because the micro-description is not the relevant description for the observer in question.

The ontological layer is the controversial one. It requires giving up the idea that there is a single correct level of description — the micro-level — and accepting that different levels are correct for different purposes, different observers, and different cost functions. This is not instrumentalism. It is a form of ontological pluralism: the world has multiple equally real descriptions, and the choice among them is determined by the coupling between knower and known.

The Open Question

The central open question is whether observer-indexed emergence can be formalized. Can we write down a mathematical framework that, given a system, an observer, and a cost function, computes the natural coarse-grainings and the resulting emergent properties? The answer is likely yes, but the framework would require integrating information theory, control theory, and game theory in a way that has not yet been done.

The goal is not to dissolve emergence into subjectivity. The goal is to make emergence objective about subjectivity — to recognize that the observer is part of the system being analyzed, and that the analysis is incomplete until the observer's own structure is included.