Interactive market lab browser demonstration — not paper results

Run a market

Choose a population, what the seller knows, how it prices, how many sellers compete, how consumers behave and which rule applies. The browser model reproduces the engine's economics but replaces seller learning with a noisy-oracle pricer, so it runs instantly; the canonical panel shows the persisted Python result for the closest configuration.

Population
Seller
Consumer behaviour (shares)
Intervention

What prices do consumers face?

Distribution of the price offered to each consumer; the vertical line is the uniform-monopoly price for this population.

Where does the surplus go?

Consumer surplus, producer surplus and deadweight loss relative to first best.

Who pays and who gains, by income?

Average price and surplus per consumer by income quintile.

How unequal is the surplus?

Lorenz curve of consumer surplus across all consumers (non-buyers at zero).

Closest canonical run paper result

Loading…

What this demo is and is not. Same valuations, signals, consumer rules, surplus accounting and policy definitions as the Python engine; but the seller here is a Bayesian noisy-oracle (posterior-mean pricing on a signal whose precision is set by the regime and quality), not a learning algorithm, competition is a differentiated-Bertrand shortcut, and explanation/data-minimization are modelled as signal degradation. Learning dynamics, audits and bandit exploration exist only in the persisted runs.