Beer Game
A four-player supply-chain simulation designed to preserve the classic Beer Game structure while allowing controlled experiments with prediction, calibrated uncertainty, factual explanations and alternative order ranges.
Why add calibrated decision support?
The classic Beer Game already shows how limited local information, delays and human reactions can create shortages, excess inventory and the bullwhip effect. Decision Lab keeps that four-role structure and information asymmetry, then varies only the model evidence available between a provisional order and the player's final order.
This lets us test a focused question: does adding a prediction, calibrated predictive uncertainty, a factual explanation or an alternative explanation change human ordering decisions — and do those local changes improve or worsen the dynamics of the whole supply chain?
The game rules and role-local information remain the same across arms; only the decision-support surface changes.
Host a session
Choose canonical or sandbox game settings, experiment arm, visible and hidden targets, interval percentiles and guard confidence. Then create four private role links.
Configure sessionJoin as a player
Players join through the private retailer, wholesaler, distributor or factory link generated by the host. Each link exposes only that role’s information.
Public descriptions and background
The Beer Distribution Game was developed at MIT Sloan as a systems-thinking and supply-chain simulation. These public sources describe the original game, its four roles, the delayed information and shipment flows, and the bullwhip effect that emerges from local decisions.
MIT Sloan's public teaching-resource description, learning objectives and background.
Public overview of the current multiplayer simulation, roles and bullwhip dynamics.
Public MIT material describing the classic board game, costs, delays and systems-learning purpose.
40 weeks · four human roles · inventory cost 0.5 · backlog cost 1.0 · two-week shipping delay · one-week order-information delay · customer demand 4 for weeks 1–4 and 8 thereafter.