Topology
In plain language: No single CEO word — “Topology” stays the research term. To a CEO it comes through its two readings, the org chart (the map) and the real wiring (the terrain), with the gap between them where intelligence gets trapped.
Definition
A Composition’s Topology is the arrangement of its constituent Agents — who is connected to whom, and how strongly. It is one attribute of the Composition, alongside the member Agents and the Constraints that bind them, and every Composition has one at every scale: a two-Agent pair has a small but real arrangement, a company a large one, with no headcount at which Topology “switches on.” It nests — a constituent Agent may itself be a Composition, so a company’s Topology has departments as its nodes, and each department has a Topology of its own. An Agent examined as a single indivisible unit has no Topology; it acquires one the moment it is examined as a Composition, the same resolution-relativity that decides whether something is treated as one Agent or many.
A Topology is a weighted graph of the couplings among its Agents. The nodes are the Agents; the edges are the couplings between them — the pathways along which one Agent’s output can reach another’s input, in place whether or not anything is flowing through them at this instant. A shared channel (a chat workspace, a hallway, the bloodstream) is not a node but part of an edge: the road, not the city.
Each edge carries a weight, and the weight is a vector, not a single number — an edge can be strong on one dimension and weak on another. A weight of zero means no pathway at all, and building one takes real work: a new channel, or, where no direct route exists, a new intermediary. Above zero, the weight is a composite whose parts live in other attributes — the channel and how much of it there is, whether the two Agents are even aware of each other (a matter of Frame), and how much the pathway actually gets used (its Realized traffic over the time-scale in question). Momentary traffic is not itself Topology; that is Actual Input and Actual Output flowing along an edge. The edge is the pathway; the traffic is what crosses it.
The same Agents sit in more than one of these graphs at once. Reporting lines are one layer, who sits near whom is another, who trusts whom a third, who actually sends work to whom a fourth — not different objects, but different edge-sets over the same nodes. Edges cross the Composition’s internal boundaries (two people in different departments who get coffee) and cross scale (one person coupled to a whole department). And a Topology is read at a frozen moment: at any instant it is fixed, and it changes over time — a reorganization, a desk move, a new channel — more slowly than the traffic running across it.
What Topology is not:
- Not the whole Composition — only its arrangement-aspect, distinct from the member Agents’ own Boundaries.
- Not the rules, and not the authority. Where a position’s permissions and authority come from is a matter of Constraints, which attach to the seat; Topology is the arrangement those Constraints operate on. Positional reach (the Positional Causal Boundary) is the pairing of the two — the seat in the Topology plus the Constraints that empower it.
- Not the momentary traffic. What is flowing right now is Actual Input and Actual Output; Topology is the structure it flows through.
- Not a single fixed layer. The org chart is one layer; the real mesh is many.
The map and the terrain
A Topology can be read two ways. The Realized Topology is how the system is actually wired — the couplings that are really there, whether or not anyone has them right. The Modeled Topology is the system’s own picture of that wiring — the model it holds of itself and runs on. In a company that picture has a familiar name: the org chart. The org chart is the map; the real wiring is the terrain; and the two come apart as a matter of course.
The gap between them is where much of an organization’s Trapped Intelligence sits. A company acts on its map — it routes decisions, information, and authority along the lines it believes are there. Wherever the terrain doesn’t match, signals that should meet never do, and capability the company already has stays stranded.
The map that matters is the system’s own model of itself — the one it acts on — not an outsider’s model of it. An outside analyst whose picture of the system is incomplete has a gap in his knowledge, not a gap in the system; and that analyst is the observer the framework leaves off the canvas (the Witness). Because the map/terrain gap needs a system that models itself, it is sharpest where that self-modeling is richest — Organizations, whose leadership keeps an explicit model of the whole. The actual wiring exists in every Composition at every scale; this particular gap rides on self-modeling and concentrates at the organizational scale.
This is a way of reading a Topology, not a separate defined term. It is the cousin, one scale up, of Frame versus Boundary — a system’s model of itself diverging from what it really is. (A different gap — the wiring a designer intended versus the one that emerged — is a target-versus-outcome question, and belongs with Target Surface, not here.)
Relations
An edge’s medium is a Topology attribute; the rules of conduct on that edge are Constraints. Protocol is the derived view of the two together — a channel plus its communication Constraints — not a primitive.
The map/terrain gap is the Organization-scale cousin of Frame versus Boundary. The neighboring gap — the wiring a designer intended versus the one that emerged — lives with Target Surface, not here. Much of Trapped Intelligence lives in the gap between the modeled and the realized Topology; the mesh is what carries a signal or strands it.
A Zoom chooses which Agents and which couplings are on the canvas; the Topology is what it chooses from. The reach a position confers — the Positional Causal Boundary — comes from where the seat sits in the Topology. And Topology is one of the Composition’s attributes; reduce the Composition to a single atomic Agent and it has none.
Example — CEO
A chief executive looks at his org chart. The lines are clean: product rolls up to the chief product officer, engineering to the chief technology officer, and the two meet only at his desk. That is the map.
The terrain is another thing. The highest-traffic line in the company is one the chart doesn’t draw at all — a senior engineer and the head of sales came up together and talk every day — and the data team gets its sharpest read on the business not from its own chain but from sitting, by accident of the floor plan, next to finance. Same people, several different wirings at once: the reporting lines, who sits near whom, who trusts whom. And the lines carry different weight — the formal link from the product manager up to the chief product officer is real but thin, used once a week; the back-channel between engineering and sales is thick, carrying most of what actually decides the roadmap.
The gap bites when a signal has to cross it. Sales knows the competitor is winning on installation speed — they hear it in every lost deal. That signal reaches engineering through the back-channel, where two friends trade notes. But it never reaches the chief product officer, who owns the roadmap, because on the terrain there is no line running there — even though the org chart says there is. The intelligence is in the building. The wiring strands it.
Example — Research
The same picture holds in a piece of heart muscle, where the parts are cells rather than people. The cells that make the heart beat are wired to their neighbors so that an electrical signal sweeps across them as a wave and they contract together. That wiring is physical: touching cells are joined by tiny tunnels between them (gap junctions), and where two cells share more tunnels the coupling is stronger, where they share fewer it is weaker — a real difference in strength on each link, exactly like the thick and thin lines in the company.
And, as in the company, the same cells carry more than one kind of wiring at once. On top of the direct electrical tunnels, the cells sit in a chemical field: signals like adrenaline diffuse through the surrounding fluid and fade with distance, so a cell near the source is coupled to it strongly and a cell far away barely at all — a link whose weight falls off, cell by cell, with distance. And they are physically glued to their neighbors, so a mechanical tug on one is felt by the others. Three different maps of who is connected to whom — electrical, chemical, mechanical — laid over the identical set of cells, each with its own strengths. That is the multi-layer mesh, in tissue instead of in an org chart.
The weights are not decoration; they carry the function. When scar tissue (fibrosis) grows between cells and wedges their tunnels apart, the electrical link between them drops toward zero — the edge is cut. The contraction wave, which needs a continuous path, breaks up and reroutes, and the beat falls out of rhythm. Nothing about the individual cells changed; the arrangement did. The heart’s problem is a Topology problem.
The company had two things — a real wiring, and a picture of that wiring it acted on, the org chart — and the whole analysis lived in the gap between them. A heart has the real wiring and nothing else. No cell, and nothing above the cells, keeps a picture of the tissue that could be right or wrong. So the map/terrain gap has nothing to attach to here because it needs something a heart doesn’t have: a system that models itself. The mesh runs all the way from a company to a heart. The gap needs a self-modeler, which is why it’s useful for the company’s analysis but not the heart’s.