Capacity

In plain language: “What’s available” or “what’s possible” — the maximum reach of the Agent’s sensing, modeling, and affecting.

Definition

Capacity is a perspective on an Agent’s Boundaries, not a separate object. It is the outermost Boundary itself — the maximum reach of the Agent’s sensing, modeling, and affecting — whether or not that reach is currently being put to work. Every Boundary definition already describes Capacity: the Computational Boundary is the Agent’s sensing capacity; the Cognitive Boundary is the Agent’s modeling capacity; the Causal Boundary is the Agent’s capacity to affect its surroundings.

Capacity becomes a named perspective when you need to compare it against what the Agent is actually doing. The companion term is Realized — what the Agent is in fact putting to work. Realized can touch Capacity but can never exceed it. The gap between them is the diagnostic space where the framework’s applied value lives: capability the Agent has but is not using.

The Capacity/Realized distinction applies uniformly to all three Boundaries and at every scale — a single Agent, a Composition of Agents, an Organization. At each scale, Capacity names the outer bound of what is possible; Realized names what is being put to work within that bound.

What Capacity is not:

• Not a separate Boundary. The framework has three Boundaries: Computational, Cognitive, Causal. Capacity is a way of reading any of them, not a fourth Boundary alongside them.
• Not fixed over time. At any analytical moment (the frozen-in-ice convention), Capacity is fixed alongside the Boundary it names. Across moments, Capacity changes when the underlying Boundary changes — through substrate change, new Tools, or a change in position within a Topology.
• Not automatically valuable to close. Whether the Agent should close the gap between Capacity and Realized depends on the Agent’s objectives, not on the gap itself. The Capacity is not Value principle governs this constraint.
• Not the Agent’s model of its own Capacity. The Agent’s actual Capacity is structurally independent of what the Agent believes its Capacity to be. Gaps between actual Capacity and the Agent’s self-model of that Capacity live at the Cognitive Boundary — specifically, at Frame. The Capacity is not Model-of-Capacity principle governs this constraint.

Relations

Capacity is the outermost Boundary — the same object defined in the Computational, Cognitive, and Causal Boundary terms, read from the perspective of what is available. Realized is the companion perspective: what the Agent is actually putting to work. The gap between them is one form of Trapped Intelligence.

Frame has a bidirectional relationship with Capacity. Frame gates Realization: the Agent’s self-model determines which of its available capabilities it actually uses. Capacity gates Frame as a ceiling: an Agent whose Frame exceeds its actual Capacity has a Frame > Boundary mismatch. A Tool the Agent has on hand extends Capacity whether or not the Agent is aware of the Tool; the Agent’s Frame determines whether the Tool is actually used.

Cognitive Capacity has a pathway for change that the other two Boundaries lack: the Cognitive Boundary can expand itself through its own modeling work. An Agent who develops a new analytical framework — or acquires one through coaching, education, or sustained practice — can analyze things he could not analyze before, without substrate change, new external Tools, or position change. The same mechanism operates when one Agent facilitates another’s Cognitive expansion: a parent helping a child learn to read, a manager coaching an employee through a new domain, a therapist reshaping a client’s model of his own situation. In each case, the Cognitive work produces new Cognitive Capacity. This is structurally distinct from Frame change (where the self-model updates but the underlying Capacity stays the same) — here, the underlying Capacity itself expands.

The Capacity/Realized gap mirrors the Frame/Boundary gap across the complexity spectrum. At both ends — trivial agents that can fully self-model, and the Infinite Agent with unbounded modeling — both gaps close to zero. In the wide middle where every Agent the framework addresses lives, both gaps are real and significant. The framework’s diagnostic value peaks in that middle.

Example — CEO

A VP has twenty years of regulatory experience, strong financial modeling skills, and deep negotiation ability. His current role uses his negotiation ability — that is what the job calls for day to day. His regulatory knowledge and financial modeling sit unused. Not because anyone decided they should not be used. Just because nobody in the organization knows he has them, or the work that crosses his desk does not call for them.

Capacity: regulatory knowledge, financial modeling, negotiation. Realized: negotiation. The gap: regulatory knowledge and financial modeling — Cognitive Capacity that is not being put to work.

The diagnostic value: the CEO is looking to hire a regulatory expert. If the VP already has that Cognitive Capacity, the answer might not be a new hire — it might be finding out why that Capacity is not being put to work. If the VP genuinely does not have the regulatory knowledge, then hiring is the right move. The Capacity/Realized distinction tells the CEO which question he is actually facing: “do we lack the capability?” or “do we have it but are not using it?”

Example — Research

A pancreatic beta cell has the enzymatic machinery to produce insulin — the genes are present, the cellular substrate can run the production pathway. That is the cell’s Causal Capacity on the insulin-production dimension.

Under normal conditions, the cell produces insulin in response to blood glucose levels. Realized output tracks Capacity closely — glucose rises, the cell secretes insulin. The gap between Capacity and Realized is small.

In early Type 2 diabetes, the cell’s substrate has not changed — the machinery is intact, the Causal Capacity to produce insulin remains. But insulin resistance in surrounding tissues and chronic overwork degrade the cell’s responsiveness. Realized output falls further below Capacity — not because the machinery is gone, but because regulatory signals and cellular stress prevent the full Capacity from being expressed. The gap widens.

The diagnostic value is the same as in the CEO case. Knowing that the Capacity is intact tells you the intervention is at the regulatory or signaling layer — the biological equivalent of Topology and Protocol — not at the substrate itself. The cell does not need new genes; it needs conditions that allow its existing Capacity to be realized. This is distinct from late-stage Type 2 diabetes, where beta cell death actually reduces Capacity. At that point, the gap closes because both Capacity and Realized have shrunk together, and the intervention shifts to substrate replacement.