Part 5 — Agent-system dynamics

How agent size and system size interact

The size of the system an agent can meaningfully operate in is related to the agent's own size — capability, coalition, and accumulated resources. Bigger agents access larger systems. But the relationship is not linear.

Theory — Geoffrey West, Scale (2017)

In biological systems, capacity scales sub-linearly (~0.75 exponent). In urban systems, innovation outputs scale super-linearly. The direction is consistent: agent scale unlocks system scale. The intensity depends on the system type. In social systems, the relationship is closer to logarithmic — doubling your size does not double your system access, but it does meaningfully expand it.

Key dynamics

Co-evolution. Agents modify the systems they inhabit. The system you play in is partly shaped by agents who played before you — and your actions change it for those who come after. This means you are never just adapting; you are also constructing. (Niche construction, Lewontin.)

Legibility cuts both ways. The system reads you as much as you read it. Selection pressure acts on what the system can perceive about you. Your signals, reputation, and legibility to other agents determine what opportunities surface and what pressure gets applied.

Structure emerges at scale. Multi-agent coalitions below a threshold can self-coordinate. Above it, they require imposed or emergent structure to function. Getting timing wrong — structure too early or too late — breaks the coalition. (Viable System Model, Beer.)

Individual optimum ≠ collective optimum. Inside any coalition, each agent has an incentive to maximize personal gain — which diverges from the coalition's interest. This is the Nash equilibrium / Pareto optimum gap. Recognizing it is the first step to designing around it.

Volatile systems punish slow adaptation. Some systems have unstable ground rules — disrupted by external systems or by agents acting at scale. In volatile systems, the speed of your feedback loop matters more than the quality of your initial strategy.

Signal and legibility

Other agents and the system itself are continuously reading you. The signals you emit — through behavior, output, affiliation, and consistency — determine what coalitions form around you and what selection pressure gets applied.

Signals that open doors

Consistent delivery on stated commitments. Legible expertise in a domain the system values. Affiliations with credible agents. Predictable behavior under pressure.

Signals that close doors

Inconsistency between stated intent and action. Affiliation with agents under expulsion pressure. Opacity about capabilities or position. Overstatement of mode (claiming influence while in survival).

Example

A founder raises a seed round, then signals aggressive growth publicly before product-market fit is secured. The signal attracts recruits and press, but also attracts the wrong investors, misaligned hires, and competitor attention — all before the agent has the energy to manage them. The legibility of the signal outpaced the actual mode. Legibility should track reality, not aspirational position.