The Wand: Influence in Persistent World Systems
Most interactive systems are built around a control assumption.
The user provides input. The system processes it. An output is returned that corresponds to what the input requested. The relationship between input and output is designed to be stable, predictable, and — in the intended case — sovereign: what the user wants is what the system produces.
This is the control model. It is appropriate for a large class of systems. It is not appropriate for persistent worlds.
The Wand is the structural formalization of a different relationship between user and system: not control, but influence.
A Formal Definition
A Wand is a mechanism for introducing external perturbation into a persistent world system.
It is distinguished from a control input by three structural properties.
State dependence. The output of a Wand action is not determined by the action alone. It is co-determined by the current state of the system. The same Wand action at Round 10 and Round 40 will produce different outcomes because the world's accumulated state is different. The Wand modifies state; the state determines what emerges.
Propagation. A Wand action does not produce a localized response. It perturbs state, and that perturbation propagates through the system according to the system's internal structure — affecting adjacent variables, modifying relationship weights, activating or suppressing unresolved tensions.
Non-sovereignty. The Wand does not dictate the outcome. It introduces a perturbation into a system that has its own causal logic. The outcome is a function of the perturbation interacting with accumulated state — not a function of the user's intent.
Direct Control vs Influence Systems
In a control system, the relationship between input and output is engineered to be direct.
A thermostat receives a target temperature. The system activates heating or cooling until the target is reached. The output is determined by the input. The system's internal state is a mechanism for achieving the input's goal — not a source of variance.
This architecture generalizes across most interactive software: text generators produce what you prompt them for, word processors output what you type, search engines retrieve results matching your query. The system's job is to realize the user's intent. Variance from intent is failure.
Influence systems operate on a different principle.
In an influence system, the user's input is not a command — it is a perturbation. The system has its own accumulated state, its own internal dynamics, its own ongoing processes. The perturbation enters this system and propagates according to the system's current configuration. The user does not determine the output; the user and the system's state jointly determine it.
The structural difference is not a matter of degree. Control and influence are different architectural commitments.
Structural Instability of Control in Persistent Worlds
Control requires a stable response space: for a given input, the output must be determinable in advance. This is possible when the system's internal state is bounded or reset between interactions.
In a persistent world, state accumulates without bound. The system at Round 40 is structurally more complex than at Round 4 — not because it has progressed, but because it contains forty rounds of accumulated causal history. The same Wand action does not merely produce a different output at different rounds; it enters a different system.
A user attempting to control a persistent world will observe a characteristic pattern: inputs that produced expected outputs in early rounds cease to do so as state accumulates. This is not system failure. It is the structural consequence of applying a control model to an influence system.
Control over a persistent world is not merely difficult. It is structurally incoherent: the stable response space that control requires does not exist. State accumulation guarantees that it cannot.
Perturbation, Propagation, Emergence
Wand influence follows a three-stage model:
Wand Action → State Perturbation → Propagation → EmergencePerturbation is the direct effect of the Wand action on the system's state. A character relationship is modified. An unresolved tension is activated. An event's causal weight changes.
Propagation is the indirect effect. A modified state variable affects adjacent variables according to the system's internal structure. A changed relationship alters how other characters respond. An activated tension affects what events become available. The perturbation does not remain local — it distributes through the system's current configuration.
Emergence is the output as rendered by the system's full state following propagation. The narrative that emerges is not the output of the Wand action in isolation — it is the output of the world's complete current state after the Wand action has propagated through it.
This is why Wand-induced outcomes are not predictable from the action alone. Accurate prediction would require knowing the action and the complete current state of the system after propagation — a state that encodes the entire causal history of everything that has occurred.
Cognitive Implications
Users trained on control systems bring a specific expectation to any interactive system: that inputs map reliably to outputs.
When this expectation encounters an influence system, a predictable failure mode emerges. The user observes that nominally similar inputs produce different outcomes at different points in the world's history. The diagnosis is inconsistency or unpredictability. The structural diagnosis is state dependence: the system is responding to its accumulated state, not failing to respond to the input.
This expectation mismatch is not resolved by making the influence system more predictable. Increased predictability would require removing the state-dependent propagation that defines it — converting an influence system into a control system, and thereby eliminating the persistent world. The mismatch is resolved by updating the cognitive model.
In an influence system, the question "what will happen if I do X?" has no stable answer. The productive question is: "how will doing X interact with the world's current state?" These questions require structurally different reasoning. The first assumes a deterministic mapping. The second assumes state dependence.
Agency in Influence Systems
The control model defines agency as sovereignty: the system does what you intend.
This definition fails in influence systems — not because agency is absent, but because the definition is category-mismatched. In an influence system, agency is real: Wand actions have genuine, measurable consequences on system state. But it is not sovereign. The outcome is a function of both the action and the accumulated state.
A more precise formulation: agency in an influence system is the capacity to introduce perturbations that alter the system's trajectory. The trajectory is not determined by the Wand alone. The perturbation is real; its downstream effects are co-determined.
This is a different model of participation, not a diminished one. Monetary policy is genuine agency: central banks cannot dictate inflation or employment outcomes, but their interventions have measurable effects on system state trajectories. Ecological management is genuine agency: introducing or removing a species alters ecosystem trajectories, even though those trajectories are not controlled by the intervention. In both cases, the agency is in the perturbation. The emergence is the system's response.
The Wand is structurally identical to these mechanisms. It is not a diminished form of control. It is the appropriate form of agency in a system that has its own dynamics.
The Wand does not write the outcome; it enters a system that was already writing itself.
For the state accumulation mechanics that determine how Wand perturbations propagate, see Game Mechanics vs State Dynamics. For the identity structure that makes Wand effects cumulative across sessions, see Instance Identity in Persistent World Systems.
