❶ The Question

What makes something a complete entity — not just a piece of information, not just a storage medium, not just a transmission channel — but something that can bear, process, and maintain information as a whole system?

Consider a neuron. It receives signals, integrates them, fires or stays silent, and modulates its own connectivity over time. Now consider the neurotransmitter molecule that crosses the synaptic cleft. The molecule carries a signal, but it does not process anything. It does not adapt. It does not maintain itself. The neuron is an operational unit; the molecule is a mechanism the neuron uses. Where, exactly, does this boundary lie? And can we define it in a way that applies not only to biological systems but to any system — physical, digital, or hypothetical — that bears and processes information?

ECI calls such an operationally coherent entity, relative to the scale of analysis, a Carrier (C). This page develops that concept.

❷ The Observation

You are a Carrier. So is a bacterium. So is an ant colony, arguably — it senses its environment, processes information collectively, maintains itself through foraging and nest construction, and adapts its behavior over time. But your DNA alone is not a Carrier. DNA holds a vast library of instructions, but it cannot read itself. It needs ribosomes to translate it, enzymes to replicate it, an entire cellular machinery to make it functional. Without the cell, DNA is an inert molecule — extraordinarily information-rich, but operationally dead.

A book is not a Carrier either. A book stores information with remarkable fidelity — some texts have survived for millennia. But a book cannot process what it contains. It cannot respond to a question, update itself when conditions change, or repair a torn page. It depends entirely on an external Carrier (a reader) to extract and act on the information it holds.

A hard drive? Same category. A neural signal? That is a process within a Carrier, not a Carrier itself. Language? A structure for encoding and transmitting information between Carriers, not an independent operational entity.

The pattern that emerges: there is a fundamental distinction between operationally coherent entities that bear, process, and maintain information, and the components, structures, and media that operate within or between them. A Carrier is the whole system. Everything else is a part of the system, a product of the system, or a mechanism the system uses to communicate.

This distinction is not always sharp — it exists on a gradient. An ant colony is a stronger Carrier candidate than a single ant, which is stronger than a single neuron, which is stronger than a single ion channel. The question is not "is this a Carrier or not?" but "how much of the Carrier function does this system embody?" ECI must define the concept clearly enough to be useful while acknowledging that nature does not always draw clean lines.

❸ What We Already Know

Several established scientific results bear directly on what it means to be an "operationally coherent entity" that processes and maintains information.

Bekenstein bound. Bekenstein (1981) proposed an upper bound on the entropy (and therefore the information content) that can be contained in a finite region of space with a finite amount of energy. For a spherical system of radius R and total energy E, the bound is:

S ≤ 2piRE / (hbar c)

where S is entropy, hbar is the reduced Planck constant, and c is the speed of light. This result arises from considerations in black hole thermodynamics and general relativity. It establishes that any bounded physical system has a finite information capacity — there is an upper limit to how much information can be packed into a given volume at a given energy.

However, a critical caveat: the Bekenstein bound is derived under specific assumptions (spherical symmetry, gravitational stability, energy conditions from general relativity). It provides an upper bound on what is physically possible. It does not directly determine the actual information capacity of any particular biological or artificial system. The information-processing capacity of a human brain, for instance, depends on neural architecture, synaptic dynamics, and metabolic constraints — not on the theoretical maximum entropy of a brain-sized region of spacetime. The Bekenstein bound tells us that capacity must be finite; it does not tell us what that capacity is.

Finite capacity in biological information processing. Decades of neuroscience research have established that biological information processing operates under severe bandwidth constraints. Individual neurons transmit information at rates on the order of 1-10 bits per second (depending on coding scheme and measurement method). Sensory systems have well-characterized channel capacities. Working memory in humans holds roughly 4 +/- 1 items. These are empirical results that confirm, independently of any theoretical bound, that biological Carriers have finite information capacity.

Autopoiesis and self-maintenance. Maturana and Varela (1973, 1980) introduced the concept of autopoiesis: a system that continuously produces and replaces its own components, thereby maintaining the network of processes that produces them. An autopoietic system is both the product and the producer of its own organization. This concept was originally developed for biological cells but has been extended to other self-maintaining systems. It provides one well-studied framework for what "operationally coherent entity" might mean: a system that maintains the very conditions required for its own continued existence.

Autonomous agents in origin-of-life research. Kauffman (2000) proposed the concept of an "autonomous agent": a system that performs thermodynamic work cycles and can reproduce. Stuart Kauffman's framework emphasizes that a living system is not merely a passive container of information but an active agent that manipulates energy to maintain and propagate its informational organization. This bridges the concepts of information processing and thermodynamic operation.

AI systems as information processors. Modern artificial intelligence systems — from simple controllers to large language models — demonstrate that information processing, learning, and adaptive behavior do not require biological substrates. A robot equipped with sensors, processors, and actuators can perceive its environment, update internal models, and act on its conclusions. These systems meet some but not all criteria that biologists associate with "operationally coherent entities" (they process information and can be autonomous, but most do not self-repair or self-replicate).

What these results collectively show: Physical systems that process information have finite capacity. Self-maintenance (autopoiesis) is one rigorously defined notion of what makes a system "complete." Information processing does not require biological substrates. But none of these results provides a single unified definition of "operationally coherent entity" that spans biology, AI, and hypothetical future systems. That is the gap ECI attempts to fill.

❹ The Framework Interpretation

ECI defines a Carrier (C) as an operationally coherent entity within a Channel that bears, processes, and maintains an Information Vector. More precisely, a Carrier is a system that satisfies four requirements:

The Four Requirements

1. Distinguishable states. The Carrier must have an internal state space with multiple distinguishable configurations. A rock at thermal equilibrium has distinguishable states in principle (positions of atoms), but its macroscopic state is effectively singular — it does not switch between meaningfully different configurations in the course of its operation. A bacterium, by contrast, cycles through states of gene expression, metabolic activity, and motility that are functionally distinguishable and consequential. The richer the state space, the greater the Carrier's capacity to represent and process information.

2. Finite capacity (K). The Carrier's information capacity K is finite. This follows from the Bekenstein bound at the fundamental level, but in practice, K is determined by the system's architecture — the number of distinguishable states, the resolution with which states can be differentiated, and the noise characteristics of the system. For a biological Carrier, K depends on neural architecture, molecular diversity, and metabolic stability. For an artificial Carrier, K depends on memory, processing bandwidth, and sensor resolution. The key point: no Carrier has infinite capacity. Every Carrier must select, compress, and prioritize. This finite capacity is not a deficiency — it is a defining structural feature that shapes how information is processed and what is possible within the system.

3. Dynamics. The Carrier must have internal processes that evolve over time — it must do something with the information it holds. A hard drive has distinguishable states and finite capacity, but it has no dynamics of its own; it sits inertly until an external system reads or writes to it. A Carrier, by contrast, actively processes information: integrating inputs, updating internal states, generating outputs. The dynamics need not be complex — a thermostat qualifies at a minimal level — but they must exist.

4. Coupling ability. The Carrier must be capable of exchanging information with its environment or with other Carriers through some Medium (see B4). A hypothetical system with rich internal states but no way to receive input or produce output would be informationally isolated — it could not participate in any larger coordination process. Coupling ability is what connects individual Carriers into networks, ecosystems, and societies. It is also what makes Carriers measurable: we can only observe a Carrier's information processing through its couplings.

Carrier vs. Components Within It

The Carrier concept draws a line between the operationally coherent entity and the components, structures, and media that exist within or between Carriers:

| Category | Examples | Why NOT a Carrier | |---|---|---| | Information structures | DNA, gene regulatory networks, language, mathematical notation | These encode information but cannot independently process or maintain it. DNA requires the cell; language requires speakers. | | Storage media | Hard drives, books, stone tablets, magnetic tape | These store information with high fidelity but have no dynamics — no processing, no self-maintenance. | | Transmission media | Neural signals, electromagnetic waves, sound waves, chemical gradients | These carry information between points but are processes or mechanisms, not entities. | | Components | Ribosomes, transistors, individual neurons (in some contexts) | These perform functions within a Carrier but do not constitute an operationally coherent entity on their own. |

Carriers: humans, bacteria, complete autonomous AI robots, ant colonies (at the collective level), individual cells, plants.

The boundary is genuinely fuzzy in some cases. Is a single neuron a Carrier or a component? Within the context of the brain, it is a component. Removed and kept alive in a dish, a neuron has some autonomous dynamics — it fires spontaneously, responds to stimuli — and might qualify as a minimal Carrier. ECI treats this as a gradient, not a binary. The Carrier concept admits degrees: a system can be more or less Carrier-like depending on how fully it satisfies the four requirements.

Carrier Containment

Every Carrier exists within a Channel:

C ∈ Ch_alpha

The Channel constrains what kinds of Carriers are possible (see B1). In our spacetime Channel Ch_ST, Carriers are three-dimensional physical systems subject to the speed of light, quantum mechanics, and the four fundamental forces. A different Channel (if any exist) might support radically different kinds of Carriers.

AI Robots as Carriers

A complete autonomous AI robot — equipped with sensors, processors, actuators, and the ability to operate independently in an environment — already satisfies the four Carrier requirements:

1. Distinguishable states: internal computational states, sensor readings, model parameters.
2. Finite capacity: bounded by memory, processing speed, and sensor resolution.
3. Dynamics: actively processes information, updates models, makes decisions.
4. Coupling ability: exchanges information with its environment through sensors and actuators.

Such a system is a Carrier now, by ECI's definition. It bears, processes, and acts on Information Vectors.

What it typically lacks is self-maintenance: the ability to repair its own hardware, replace degraded components, or reproduce. This does not disqualify it as a Carrier — self-maintenance is not one of the four requirements. But it does place it in a different regime from biological Carriers. A biological Carrier that self-maintains, self-repairs, and self-replicates occupies what ECI calls the life-like regime (see D1). An AI robot that acquires these capabilities — repairing its own circuits, fabricating replacement parts, building copies of itself — would cross into that regime. The Carrier concept is substrate-independent; whether a system is "alive" depends on what it does with its Carrier capabilities, not on what it is made of.

The Boundary Question

When is a system "complete enough" to count as a Carrier? ECI proposes that this is a gradient, not a binary, and suggests measuring along the four dimensions:

• State richness: How many meaningfully distinguishable states does the system have?
• Capacity utilization: How close does the system operate to its capacity limits, and how effectively does it manage finite resources?
• Dynamic complexity: How elaborately does the system process information — simple stimulus-response, or complex internal modeling?
• Coupling range: How many types of information exchange can the system engage in, and how far does its influence reach?

A thermostat scores low on all four. A bacterium scores moderately. A human brain scores high. An ant colony scores differently depending on which level you analyze — individual ants are modest Carriers; the colony as a whole may be a more capable one. These are empirical questions, not definitional ones.

❺ If This Were True...

If the Carrier concept — substrate-independent, defined by operational capabilities rather than material composition — is the right way to carve nature at its joints, several consequences follow.

Non-biological Carriers are not second-class citizens. An AI system that satisfies the four requirements is as much a Carrier as a bacterium or a human. This is not a metaphor or an honorary title — it follows directly from the definition. If we eventually build systems that also self-maintain and self-replicate, they would enter the life-like regime. The question "is an AI alive?" becomes a question about capabilities, not about substrate.

Carrier-independent information science becomes possible. If the Carrier concept is well-defined, then we can study information dynamics — how information is borne, processed, maintained, and exchanged — without restricting ourselves to one type of Carrier. The same coordination principles (see Variation & Coordination, C1) might apply to cells, organisms, AI systems, and hybrid bio-digital networks. This would unify currently fragmented fields: neuroscience, artificial intelligence, ecology, and origin-of-life research all study Carriers, but they rarely share theoretical frameworks.

The Carrier boundary problem becomes empirically tractable. Rather than debating whether viruses are "alive" or whether large language models are "conscious," we can ask: where does this system fall on the four-dimensional gradient of Carrier capabilities? This reframes philosophical debates as measurement problems — which, even if they remain difficult, are at least approachable with empirical tools.

Capacity limits shape everything. If every Carrier has finite capacity K, then the way a Carrier selects, compresses, and prioritizes information is not incidental — it is a fundamental constraint that shapes the Carrier's behavior, its coordination with other Carriers, and the evolutionary pressures it faces. This connects to the observer and compression discussions elsewhere in the ontology (see Observer & Compression, D3b).

These are extrapolations, not predictions. They sketch the intellectual landscape that opens up if the Carrier concept proves useful.

❻ How Could We Test It?

The Carrier concept can be tested at two levels: whether the definition is useful (does it carve nature at a productive joint?) and whether it is predictive (does it generate testable claims?).

Cross-system comparison. Select a diverse set of candidate Carriers: bacteria, nematodes, octopuses, humans, robot controllers, neural network agents, ant colonies, slime molds. For each system, independently measure the four Carrier dimensions (state richness, capacity, dynamics, coupling ability). Then measure the system's information-processing performance: adaptability, error correction, coordination with other systems, persistence under perturbation. If the four-dimensional Carrier profile predicts information-processing performance better than simpler variables (system size, metabolic rate, number of components), the Carrier concept adds explanatory power.

Capacity-behavior relationships. If finite capacity K is a structurally important feature, then systems operating near their capacity limits should exhibit qualitatively different behavior from systems with spare capacity — more compression, more selective attention, more information loss under load. This is testable in both biological systems (cognitive load experiments in humans, information bottleneck analysis in sensory systems) and artificial systems (neural networks approaching memory or bandwidth limits).

Carrier boundary experiments. Construct artificial systems with varying degrees of Carrier completeness — from purely passive storage (hard drive), to storage with simple dynamics (cellular automaton), to full autonomous agents. Test whether there is a meaningful transition in information-processing capabilities as systems become more Carrier-like along the four dimensions. If the transition is gradual but exhibits critical thresholds (e.g., the system's ability to maintain information patterns jumps sharply when coupling ability is added), this supports the Carrier concept as capturing something real about the organization of information-processing systems.

AI Carrier experiments. Compare AI systems with and without self-maintenance capabilities. If adding self-maintenance (automatic error correction, component replacement, resource management) produces a qualitative shift in information dynamics — not just improved reliability but new capabilities such as open-ended adaptation — this supports the distinction between "Carrier" and "life-like Carrier" that ECI proposes.

What would weaken this claim: If the four requirements do not correlate with information-processing capabilities — i.e., if systems satisfying fewer requirements perform just as well as systems satisfying more.

What would kill this claim: If no meaningful distinction can be drawn between "operationally coherent entities" and "components/media" — i.e., if every level of organization is equally well described as a Carrier, making the concept vacuous.

❼ Connected Nodes

→ ECI Unit: The minimal dynamic unit, composed of Information, Carrier, and Energy. The Carrier is the "C" in ECI — the entity that bears and processes Information Vectors using Energy.

→ Channel & Dimensional Architecture (B1): The arena within which Carriers exist. Every Carrier is embedded in a Channel (C ∈ Ch_alpha), and the Channel constrains what kinds of Carriers can form and how they can operate.

→ Coupling & Resonance (B3): How Carriers interact through information. Coupling ability is one of the four Carrier requirements — without it, a Carrier is informationally isolated and cannot participate in coordination.

→ Medium (B4): The mechanisms through which Carriers exchange information. Medium is distinct from Carrier: the Carrier is the entity; the Medium is the mechanism of exchange between entities.

→ Life (D1): What happens when a Carrier acquires self-maintenance, self-repair, and self-replication. The Carrier concept defines the broad category; Life defines a specific regime within it — the regime where Carriers begin to maintain the conditions of their own persistence.

❽ Mathematical Detail

Definition: Carrier (C)

A Carrier is a system that satisfies the following requirements:

1. State space: C possesses a state space S(C) with |S| > 1 distinguishable states.
2. Finite capacity: The information capacity K(C) is finite: 0 < K(C) < infinity.
3. Dynamics: There exists a time-evolution operator T such that the state s(t) of C changes over time: s(t + delta_t) = T(s(t), inputs).
4. Coupling: C can exchange information with at least one other system or environment through some coupling function gamma.

• Status: Definition (framework notation).
• Assumptions: That "distinguishable states," "finite capacity," "dynamics," and "coupling" can be operationalized for any physical or computational system; that these four requirements jointly define a meaningful category.

Containment:

C ∈ Ch_alpha

Every Carrier exists within a Channel Ch_alpha. The Channel constrains the Carrier's physical embedding dimension (D_embed(C) ≤ D_spatial(Ch_alpha)), causal structure, and available interactions (see B1 for details).

• Status: Definition (framework notation).

Capacity (K):

K(C) = max information content the Carrier can maintain

The capacity K is bounded above by the Bekenstein bound for any spatially bounded system (under the bound's assumptions), but in practice K is determined by the system's architecture:

• For biological Carriers: K depends on neural/molecular architecture, metabolic constraints, and noise characteristics.

• For artificial Carriers: K depends on memory, bandwidth, and sensor resolution.

• For all Carriers: K is finite and determines the system's fundamental information-processing constraints.

• Status: Proposed (the concept of a measurable Carrier capacity is a framework claim, not an established result).

• Assumptions: That information capacity can be meaningfully defined and measured for diverse physical systems.

• Falsifiable consequence: If K cannot be operationalized — if there is no meaningful way to measure or compare information capacity across different types of Carriers — the concept fails.

Carrier-Vector relation:

C bears 𝐈 (Information Vector)

A Carrier bears, processes, and maintains one or more Information Vectors (see A2). The Carrier provides the physical substrate and dynamics; the Information Vector is the organized information pattern that persists within the Carrier. The same Information Vector could, in principle, be borne by different Carriers (substrate independence) — though this claim is stronger than the basic definition and requires separate justification.

• Status: Proposed.
• Assumptions: That the Carrier-Vector distinction maps onto a real structural feature of information-processing systems, not merely a notational convenience.
• Falsifiable consequence: If information patterns cannot be meaningfully separated from their physical instantiation — if "the same information" in two different substrates is not a coherent concept — then the Carrier-Vector distinction collapses.

Carrier gradient measure (sketch):

The "Carrier-ness" of a system could potentially be quantified along the four dimensions:

• sigma_S: state richness (number and distinguishability of available states)
• sigma_K: capacity utilization (how effectively finite capacity is managed)
• sigma_T: dynamic complexity (elaborateness of internal information processing)
• sigma_gamma: coupling range (diversity and reach of information exchange)

A composite Carrier index might take the form:

Phi_car = f(sigma_S, sigma_K, sigma_T, sigma_gamma)

where f is a function to be determined empirically. This is a sketch, not a formal proposal — defining f and validating the component measures is a priority for future mathematical development. The key prediction: Phi_car should correlate with independently measured information-processing performance.

• Status: Speculative sketch.
• Variables: sigma_S, sigma_K, sigma_T, sigma_gamma are not yet formally defined.
• Falsifiable consequence: If the four dimensions do not independently contribute to predicting information-processing capability, the gradient model is overcomplicated and should be simplified.