Simplified Simulation¶
Software or tool — instantiates Bounded Approximation
Simulates a reduced version of the system that captures enough behavior to guide the decision.
A Simplified Simulation is a runnable model of a deliberately reduced version of the system — fewer actors, collapsed states, coarser time, linearized or averaged dynamics — chosen so that the structure removed is not the structure the decision depends on. Its defining discipline is honesty about the cut: you say what was stripped out, and you confirm that what was stripped is not decision-critical. Unlike a surrogate, which only learns an input-output mapping of another model, a simplified simulation still reproduces the system's mechanism at reduced fidelity, so it can show you emergent behaviour — queues forming, feedback building, a tipping point — that no static estimate would surface. It is a small dynamical world built to answer one question the real world is too expensive to run.
Example¶
A city is deciding whether to add a dedicated left-turn signal phase at a congested intersection. A full microsimulation of the surrounding road network is heavy to build and calibrate, and the decision does not need it. So the traffic team builds a simplified queue simulation of just that junction and its two feeder roads: fixed arrival rates by hour, no pedestrians, no rare stalled-vehicle incidents, cars as identical units. They run it under the current signal timing and under the proposed left-turn phase and watch the queue lengths. The reduced model shows the new phase clearing the eastbound backup without starving the cross street — enough to justify a field trial. What was cut (pedestrians, incidents) is named explicitly, and the team checks that none of it is what drives peak-hour queueing before trusting the result.
How it works¶
- Fix the decision-critical behaviour. Name the specific dynamic the decision hinges on — here, peak queue length under a timing change.
- Strip the rest. Remove actors, collapse states, and coarsen time resolution wherever doing so does not touch that behaviour.
- Encode reduced dynamics. Implement the simplified mechanism as a runnable model.
- Run and read. Execute across the relevant conditions and read off the behaviour that maps to the decision.
- Confirm the cut was safe. Check that restoring an omitted element would not change the conclusion.
Tuning parameters¶
- Fidelity level — which dynamics are retained versus abstracted. Higher fidelity captures more real behaviour at more modelling and runtime cost, and more ways to be wrong.
- Time resolution — how finely time is stepped. Coarser steps run faster but can smooth over short-lived surges the decision may care about.
- Entity and state granularity — how many distinct actors and states the model tracks. Collapsing them speeds the run but can hide heterogeneity that matters.
- Stochastic vs deterministic — whether randomness is modelled. Deterministic runs are cheaper and clearer; stochastic runs reveal variability and tail behaviour.
When it helps, and when it misleads¶
Its strength is revealing dynamic, emergent behaviour — congestion, oscillation, cascades, thresholds — at a fraction of the cost of a full-fidelity model, letting a team see how the system behaves, not just how big a number is. The honest posture behind it is George Box's dictum that "all models are wrong, but some are useful": the reduction is justified by fitness for the decision, not by realism.[n1]
Its failure mode is that the removed structure turns out to be decision-critical after all — an ignored actor, a rare incident, a feedback loop that only bites at scale — and the simulation smooths the very effect that should have driven the choice (threshold blindness). The classic misuse is showing a stripped-down simulation's polished animation to stakeholders as if it were the real system. The guarding discipline is to keep the list of cuts visible and to test whether restoring any single cut changes the conclusion — a job handed to Sensitivity Probe.
How it implements the components¶
approximation_method— the reduction (remove actors, collapse states, coarsen time) is the simplification technique.assumption_scope— the explicit list of what was removed, held constant, linearized, or averaged.validity_domain— the operating conditions (arrival ranges, no major incidents) under which the reduced model mirrors reality.decision_requirement— the specific behaviour the simulation must reproduce well enough, which sets what may safely be cut.
It does not run a validation_check against real-world outcomes — Policy Pilot and Surrogate Model do — nor state a formal acceptable_error bound the way Rough Order-of-Magnitude Estimate does.
Related¶
- Instantiates: Bounded Approximation — a reduced dynamical model standing in for the full system, bounded by stated cuts and validity conditions.
- Sibling mechanisms: Back-of-Envelope Estimate · Rough Order-of-Magnitude Estimate · Surrogate Model · Algorithmic Relaxation · Prototype Test · Policy Pilot · Sensitivity Probe
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Simplified Simulation operates as an analytical, modeling, inference, comparison, or optimization procedure that derives insight or a solution because it simulates a reduced version of the system that captures enough behavior to guide the decision.
Independent corroboration: The frozen evidence defines Simplified Simulation as 'Simulates a reduced version of the system that captures enough behavior to guide the decision', so its operative form is Analysis, Modeling & Optimization.
Nearest alternative: Experiment, Test & Rehearsal — Simplified Simulation includes features of an active test, trial, simulation, drill, or rehearsal that generates evidence through a deliberate attempt or perturbation, but its defining operation is an analytical, modeling, inference, comparison, or optimization procedure that derives insight or a solution.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Systems Thinking & Cybernetics
Origin pattern: Convergent development
Present-day reach: Universal
Rationale: A reduced model that retains decision-relevant dynamics is a systems-modeling abstraction. NASA simulation credibility guidance distinguishes purpose-adequate model fidelity from exhaustive representation; computing provides execution.
Related originating lineages:
- Computer Science & Software Engineering — computer_science contributes computer science and software-engineering practice to this mechanism's defining operation—Simulates a reduced version of the system that captures enough behavior to guide the decision—without displacing the selected primary historical lineage.
- Engineering & Design — Simplified simulations support early trade studies before high-fidelity models are warranted.
- Operations Research — Decision models deliberately abstract the system to relevant variables and constraints.
- Statistics & Experimental Design — Validation tests whether reduced behavior is adequate for the intended inference.
Review resolution: The blind reviewers disagree on primary lineage (systems_cybernetics versus computer_science). Authoritative or primary research supports systems_cybernetics as the best historical origin: A reduced model that retains decision-relevant dynamics is a systems-modeling abstraction. NASA simulation credibility guidance distinguishes purpose-adequate model fidelity from exhaustive representation; computing provides execution. The cited NASA Software Engineering Handbook, Models and Simulations; NASA, Simulation Credibility Guide directly supports the mechanism's defining operation. All independently supported contributing domains are retained without an arbitrary cap. origin_mode=convergent records lineage, while domain_reach=universal records later applicability separately from provenance.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Researched adjudication after independent review; high confidence.
Sources consulted:
Notes¶
[n1] Statistician George E. P. Box's aphorism, "All models are wrong, but some are useful," is the working creed of simplified simulation: a model earns its keep by being adequate for a particular decision, so the right question is never "is it realistic?" but "is what I left out able to change the answer?" ↩