Composite Actors

The individual actor is not the only useful unit of prediction, because the boundary of an actor is chosen relative to the transition and trace under study; an ant moves under local chemical gradients, contact, hunger, alarm, and whatever else an ant can register, while the colony moves under distributed paths, food reserves, brood state, nest condition, caste, traffic, and the aggregate consequence of many local actors whose behavior contains both stable pattern and randomness. The ant is an actor; the colony is also an actor; neither statement cancels the other.

The same principle extends into institutions. A corporation contains people acting with varying independence, yet the corporation also persists, remembers, receives information, filters it, makes decisions, and changes the world under one operational boundary; an employee sends an email, a committee rejects a contract, the corporation exits a market, and these are actions at different scales produced by actor constructions nested inside one another.

The outer grammar is repeated here because the corporate derivation depends on it.¹

An Actor Is a Choice of Scale

The useful actor is selected relative to the trace and decision being modeled. If the outcome is whether one executive replies, model the executive inside the institution; if the outcome is whether the company acquires another company, the executive becomes one component inside a corporate actor; if the outcome is a market transition, several corporations may become components of a larger system whose state cannot be reduced to any one member.

Whatever can be traced at a declared level of abstraction may be treated as the action of the actor constructed at that level, provided the construction has a persistent boundary, selective access to information, an organization through which information becomes consequential, memory, and repeatable outward transition. Statistical predictability alone is not enough, because a random collection of correlated things is not thereby one actor; the boundary and transition grammar must explain why the components jointly produce the trace.

The Organization Is a Synthetic Object Before It Is a Composite Actor

An organization exists first as a synthetic construction inside the realities of the people who participate in, observe, fear, regulate, exploit, or depend on it. “The company” is not one object copied into every mind; a founder, employee, child, creditor, regulator, or person with a radically different affective structure may construct entirely different organization-objects from the same legal entity, because each receives different information, assigns different salience, and possesses a different relationship to authority, obligation, danger, and belonging.

The institution nevertheless becomes more than a collection of private feelings when those actor-relative constructions are coupled through records, routines, communication channels, authority, incentives, assets, laws, and recurring consequences; the people act through what they believe the organization is, those actions alter the shared synthetic structure, and the resulting pattern persists long enough to confront later propositions as one higher-order actor. The corporation is therefore both a phenomenal object inside individuals and a composite phenomenal actor at another scale, where phenomenal means the organized reality made available to the composite through its members and institutional machinery, not a declaration that the corporation contains one human-like stream of consciousness.

Institutional Access and Interpretation

Part 1 described an actor as an access map plus an interpretive organization; a corporation follows from the same principle, because a fact cannot influence a corporate decision merely by existing somewhere in the building, it must reach the relevant institutional channel and survive whatever authority, delay, incentive, and distortion governs its path.

Let \(A_{C,t}^{\mathrm{inst}}\) denote the corporation’s institutional access architecture: what reaches the organization, through which channel, with what delay, after which filters, and under whose permissions. A fact known by an intern but blocked from the decision-maker may exist in the corporation’s environment without entering the state relevant to the decision; a dashboard, board packet, customer escalation, legal opinion, rumor, and private conversation create different paths of access even when they concern the same external condition.

Let \(\mathcal I_{C,t}^{\mathrm{inst}}\) denote the corporation’s institutional interpretive organization: governance, incentives, authority, procedure, culture, coalition structure, veto rights, active objectives, and the formal or tacit rules by which information becomes action.

GIDS–15 — Institution-relative object construction.

Read it exactly as the human equation. Something happens outside the institution; institutional channels make some part of it available; the existing corporate state and organization determine what the event becomes for the company. A regulatory notice may become an existential threat, an ordinary compliance task, an opportunity to crush smaller competitors, or nothing at all because it died in the wrong inbox; the document is the same, while the institutional object is not.

This construction makes the composite actor depend on the philosophy rather than appearing later as a convenient aggregation box. Information architecture is the institutional access map; governance, incentive, and authority are the interpretive organization.

Constructing the Corporate State

Let \(\mathcal J_C(t)\) be the people relevant to corporation \(C\) at time \(t\), and let \(\varsigma_{j,t}^{\mathrm{person}}=(T_{j,t},\phi_{j,t},c_{j,t})\) be the conceptual person-state contributed by member \(j\). The corporate construction must also contain facts and statistics, authority and communication structure, institutional memory, incentives and active constraints, and environmental history already absorbed by the institution.

GIDS–16 — Composite institutional state.

The aggregator \(\mathcal A_{\mathrm{corp}}\) is not an average, it is the formal interface for the institutional access and interpretation process just described; it must represent unequal authority, missing information, coalitions, vetoes, delegated action, procedural delay, and the fact that most employees contribute nothing to most decisions. The equation does not explain those mechanisms by itself, although it prevents an implementation from quietly replacing them with mean-pooling and continuing to call the result a corporation.

The state is not limited to formal decisions. Any persistent institutional change may matter: a new internal belief, changed risk posture, budget reallocation, relationship fracture, new champion, lost executive, policy, hidden queue of work, or market action; what counts depends on the scale of the actor and the traces available.

A corporation acquires an almost feel-like way of dealing with the world, not necessarily human consciousness but a characteristic pattern of access, memory, salience, delay, and response. Some companies absorb bad news early and act; some route it upward until it has been sanded into harmlessness; some experience every external proposition through one founder; some possess no center until crisis supplies one. That pattern is part of the corporate phenomenal state at the scale we have chosen.

Partial Observation Is a Projection Problem

The complete corporate state is never observed. We may see three people in a buying committee of eight, one official org chart, stale funding data, a few public statements, and the traces produced in our own relationship; the missingness is structural, therefore it belongs in the state estimate instead of being silently converted into zero.

Let \(\mathcal J_C^{\mathrm{obs}}(t)\subseteq\mathcal J_C(t)\) be the members we can use, let \(\mathbf u_{C,t}^{\mathrm{corp}}\) collect measurable institutional evidence, and let \(\mathbf m_{C,t}^{\mathrm{miss}}\) encode missing membership, authority, and fields. The estimate is

A missing executive is not a zero-vector executive, unknown authority is not equal authority, and an unseen veto player should widen uncertainty rather than disappear. Conceptually, observability is a projection from complete institutional state into available traces,

while the estimator attempts an inverse problem under uncertainty and never literally reconstructs everything the projection destroyed.

Relationship as a State of Its Own

Two actors create structure belonging to neither actor alone. Let \(\Gamma_{ab,t}\) denote their relationship—familiarity, trust, perceived authority, prior commitments, objections, channel history, response rhythm, resentment, debt, affection, status, and whatever else the interaction has accumulated.

A relationship is not merely actor \(a\)‘s opinion of actor \(b\) plus actor \(b\)‘s opinion of actor \(a\); it contains a sequence of mutual transitions, because one promise changes the meaning of the next promise, one betrayal changes the meaning of silence, and the relationship becomes an object precisely because it persists and alters future response.

Estimate it from pre-proposition history,

then update it when relevant records arrive. Chronology matters; a later apology cannot leak backward into the relationship-state used to predict the insult.

The Dyadic Actor

The first practical world model uses a dyad. Let \(a\) be a sender or initiator, \(b\) the receiving actor, and \(C_a,C_b\) the institutions through which they act.

GIDS–17 — Filtered dyadic state.

The dyad contains two people, two institutional contexts, one relationship, and a shared world; a commercial interaction is one laboratory, while the same state type can represent negotiation, recruiting, diplomacy, command, care, instruction, or any repeated proposition-response process involving two sides.

There is deliberate redundancy, because the focal people may already contribute to their company representations; an implementation must either exclude them from the corresponding company-context summary or estimate the coupled states jointly, otherwise the same evidence is counted twice and called confidence.

A proposition may contain content, offer, framing, evidence, channel, timing, sender, and requested action, although this is not a universal schema; the proposition could instead be a policy, threat, candidate, medical intervention, educational sequence, or arrangement of physical conditions. What matters is that it enters the actor through an actor-relative representation.

The dyad is not a new metaphysical species, it is the convenient actor construction at the scale where relationship, institutions, and proposition jointly determine the next trace. The internal construction changes with the actor, while the outer algebra remains stable; that is how realities become composable without becoming the same.

The next chapter takes this composite state and does the dangerous thing: it introduces candidate propositions, simulates what follows, then asks which future should be preferred.