Actual Input

In plain language: “What’s actually reaching you right now.”

Definition

Actual Input is the specific information that is in fact crossing an Agent’s Computational Boundary at a given position and time. It is a property of (Agent × position × time) — not of the Agent alone. The same Agent in the same company on the same day receives different Actual Input depending on where he sits in the Topology, what meetings he is in, what emails have arrived, and what conversations he happens to walk past.

Actual Input is the time-and-context-dependent counterpart to the Computational Boundary, which is treated as time-and-context-free. The Computational Boundary is the channel surface — what the Agent could receive. Actual Input is what the Agent is receiving. The Boundary is the capacity; Actual Input is the flow.

What Actual Input is not:

• Not the Computational Boundary itself. The Boundary is the Agent’s channel surface, independent of what is flowing through it. Actual Input is what is flowing. A channel that exists but carries nothing (a VP who has an open-door policy but nobody walks in) is part of the Computational Boundary at Capacity with zero Actual Input through that channel.
• Not what the Agent does with what it receives — that is the Cognitive Boundary (modeling) and ultimately the Causal Boundary (acting). Actual Input stops at the perimeter; what happens after receipt is a different Boundary’s territory.
• Not fixed. Actual Input changes as the Agent’s position in the Topology changes, as time passes, and as the environment shifts. This is the object in the framework that carries context-dependence — the Boundaries are frozen, and Actual Input is where “right now” lives.

Relations

Actual Input sits between the Computational Boundary and Realized on that Boundary. The Computational Boundary is the channel surface the Agent has by virtue of being that Agent. Actual Input is the specific signal crossing that surface at a given position and time. Realized on the Computational Boundary is the broader perspective on what is actually happening with that Boundary in a given context — Actual Input is the Computational Boundary’s specific form of Realized. The gap between Computational Boundary and Actual Input is one named form of Trapped Intelligence: capacity the Agent could use if the signal reached it, but the signal doesn’t. The Agent’s Topology position determines which subset of the world’s signals actually cross the Boundary — the same Agent at a different node in the org chart receives different Actual Input even though the Computational Boundary is unchanged. Actual Output is the Causal-side symmetric term.

Example — CEO

A VP of sales has a Computational Boundary that includes face-to-face conversation, email, phone calls, and video conferences — these are his channels. His Actual Input on a Tuesday morning is: a pipeline report that arrived at 8am, a Slack thread about a delayed deal, and a hallway conversation with his CTO about a product limitation. By Thursday afternoon his Actual Input is different — a customer escalation email, a board-prep document, and a one-on-one with a rep who is about to resign. The Computational Boundary didn’t change. What flowed through it did.

The diagnostic value is in the gap. The VP’s Computational Boundary includes the ability to receive field reports from regional offices — the channel exists. But if the Topology routes those reports through a regional director who summarizes them before passing them up, the VP’s Actual Input is the summary, not the raw reports (for better or worse). Information the VP could use — a pattern visible only in the raw data — never crosses his Boundary in practice. The capacity is there; the Topology is blocking the flow. This could be Trapped Intelligence at the Computational layer if the summaries are inadequate: the Agent’s channels are there, but the signal that makes a difference isn’t reaching them. Notably, the opposite could also be true: overly detailed reports might have all the right information, but buried in a quantity the VP can’t process — so that is also Trapped Intelligence, just at the Cognitive layer rather than the Computational one.

Example — Research

A biological cell’s Actual Input is the specific set of signals binding to its receptors at a given moment in a given tissue location. The cell’s Computational Boundary includes all receptor types it expresses — its full channel surface. But at any given instant, only a subset of those receptors are actively binding ligands. Which subset depends on what molecules are in the cell’s immediate environment, which depends on where the cell sits in the tissue (its position in the Topology) and what neighboring cells are secreting (their Actual Output).

Move the same cell from inflamed tissue to healthy tissue and its Computational Boundary is unchanged (same receptors), but its Actual Input shifts dramatically — different cytokines, different concentrations, different signaling context. The cell’s Cognitive Boundary (its signaling cascades) responds to whatever is actually binding, not to what could theoretically bind. Actual Input is the bridge between what the cell can detect and what the cell is detecting.