Trials

One world, every contingency.

A trial is a run of a world with its parameters set. Sweep a parameter and every outcome is a run you can open, compare and replay.

A geoeconomics simulation on a globe in the CommandAGI workbench: oil flows from the Persian Gulf to China, India, Japan, South Korea, Europe and the United States drawn as arcs, each country coloured by its oil inflow, while the globe turns slowly from the Gulf toward the Pacific.
Recorded in the workbench: the geo-hormuz world on its globe. Its numbers are illustrative, not data.

A sweep

Plan the trials, run each one.

The trial planner reads the parameters and their bounds from the world's file. Each trial is a simulated run whose simulator.settings hold its values, so it can be run again exactly.

How a sweep of trials runsThe world file declares each parameter with its bounds. The trial planner makes one setting per value. The engine steps each trial for its horizon, and each one's outcome is kept with the values that made it, so any trial can be run again from its settings.world.tsxescalation: 0 to 3, step 0.1The trial plan11 values of escalation11 simulated runssimulator.settings:param.escalationOutcomespeak price, closure, stocksHow a sweep of trials runsThe world file declares each parameter with its bounds. The trial planner makes one setting per value. The engine steps each trial for its horizon, and each one's outcome is kept with the values that made it, so any trial can be run again from its settings.world.tsxescalation: 0 to 3, step 0.1The trial plan11 values of escalation11 simulated runssimulator.settings: param.escalationOutcomespeak price, closure, stocks

Eleven trials, 120 days each.

scripts/sites/case-studies.mjs hormuz swept Iran's escalation from 0 to 3 in steps of 0.3 in the one engine and kept each trial's outcome in that file.

case-studies/hormuz/stats.json · trialsthe engine's outcomes · simulation · 2026-11-04
escalation 0.0–0.9  peak  $80.0          closure  0.0%  India 70.0  China 900.0
escalation 1.2      peak  $90.5  day 74  closure 12.4%  India 64.8  China 883.5
escalation 1.5      peak $103.9  day 74  closure 28.3%  India 57.2  China 859.3
escalation 1.8      peak $117.3  day 74  closure 44.2%  India 49.0  China 833.4
escalation 2.1      peak $130.7  day 74  closure 60.2%  India 40.5  China 806.5
escalation 2.4      peak $144.1  day 74  closure 76.1%  India 31.8  China 779.0
escalation 2.7      peak $155.8  day 74  closure 90.0%  India 23.5  China 752.7
escalation 3.0      peak $155.9  day 77  closure 90.0%  India 20.9  China 744.5

What the trials say

Nothing happens below 1.2.

Iran's effort must pass 1.5 before the strait closes at all. Past 2.7 the closure is at its cap of 90%, and the peak price stops rising.

A trial is a run. Each one is recorded beside the world, with reality: simulation, the principal that started it and its settings. Same file, same numbers. A trial saved as a world steps to the same values as the trial it came from. A test checks it. Read the sheet. A studio's trials can take their values from a sheet, so a person or an agent can list the cases in a table.

trials of 120 days
11
peak price up to escalation 0.9
$80.0
peak price at escalation 3
$155.9

Simulated

Peak oil price by escalation, USD a barrel
Les chiffres
Peak oil price by escalation, USD a barrel
EscalationPeak price
080
0.380
0.680
0.980
1.290.5
1.5104
1.8117
2.1131
2.4144
2.7156
3156
ProvenanceFile cloud://global/examples/template-library/worlds/geo-hormuz/world.tsxOutput case-studies/hormuz/stats.jsonTool scripts/sites/case-studies.mjs hormuz: sim-engine.jsRan on This computer (dell-gb10-1)When

Sweep a parameter of your own.

Any number a world declares as a parameter can be swept.