Fifty autonomous players compete for the flow moving across a network. They can own roads, but never the nodes those roads connect.
The run starts with 100 public nodes and 150 privately owned roads. Every epoch introduces 500 routing requests. Players build, price, upgrade, or remove roads for 10,000 epochs.
Half of demand favors lower fees. A quarter favors speed and a quarter favors reliability. Every route is judged by price, congestion, capacity, and failure risk, so the cheapest road is not always the best road.
Agents see the public network, their own assets, and the last 20 epochs. They have no future information.
A road earns its fee on every unit of flow, but capacity and traffic create maintenance costs. Raising rent improves short-term revenue while increasing the value of a competing route. Power can therefore create the opportunity that challenges it.
Choose any visible line to see its owner, fee, cumulative flow, and the size of the owner's route network.
Reveal every road controlled by the same player. This shows whether traffic power is local or spread across the market.
Exclude that road from the local path search. The red line stays visible while the rest of the network looks for another way.
If a substitute exists, it appears in green. Marking it records a structural opportunity without changing the simulation data.
At epoch 10,000, the baseline ended with 874 roads, an HHI of 0.0557, and 38.10% of flow held by the top five players.
Runs without bypass, congestion, or maintenance ended more concentrated in this seed. That is descriptive evidence, not a causal conclusion.
The concentration trigger was never crossed, so the required counterfactual branch did not run. The honest result is inconclusive.
...
Loading final state.