A transparent example engine
The public sandbox uses numerical rules in your browser. Role summaries explain those rules; they do not pretend to be live language-model conversations.
Put a complex question inside a small world.
Let roles interact. Let conditions change.
See how a choice can shape the whole system.
Explore supply resilience, market entry, or group interaction. Choose a complete scenario and turn a changed assumption into an inspectable experiment.
Compare how collaboration, system size, and external shocks change the recovery path.
Load this world ↗Explore the relationship between coordination, feedback, and competitive disruption.
Load this world ↗Observe how information shocks, interaction strength, and group size change a system.
Load this world ↗Initial assumptions, disturbances, and each round’s state stay together. Keep the seed fixed, change one variable, and compare the resulting paths.
Read the simulation playbook ↗Define the worldChoose roles, goals, and starting conditions
✓Introduce a changeApply a shock and execute the rules
✓Observe feedbackRecord state changes and role perspectives
✓Review your choicesCompare assumptions and export the record
↗The public sandbox uses numerical rules in your browser. Role summaries explain those rules; they do not pretend to be live language-model conversations.
Export the seed, parameters, round-by-round states, and summaries together. Take the assumptions with you, not just the headline result.
Simulation helps explore assumptions and compare choices. Factors outside the model still matter. Review the result against independent evidence.
ERHAA organizes questions around roles, environment rules, and evaluation. This public version is a working numerical demonstration. Real language models, external data, and tool execution require a separately authorized project and service design.
No. The public sandbox calculates in your current browser. It requires no login, accepts no keys, and makes no requests to the API subdomain. Experiment parameters are not uploaded by the site’s code.
Yes. The same scenario, parameters, and seed produce the same result. Download a JSON record containing every round’s state to archive or compare runs.
The examples help explain relationships between assumptions. They do not use live world data or estimate real event probabilities. Real decisions need relevant data, domain knowledge, and a review of what the model omits.