More ride requests do not always produce more completed trips. When a platform serves the oldest waiting passenger first across a very large area, the next passenger may be far from every free vehicle. Cars spend more time driving empty to pickups, leaving less time to carry people.
The first simulation shows ordinary matching, pickup, occupied travel, and return to availability. At much higher request volume, pickup paths grow long and service completion can fall despite stronger demand. The resulting curve rises and then bends downward, the wild-goose-chase effect. A final run divides the area into smaller matching zones, shortening empty pickups and improving service in the illustrated case. The lesson is that dispatch rules and market geography determine usable capacity. Sending a driver after every request is not enough; long approach trips can consume the supply that should have served the next passenger.
An available taxi is useful to passengers even while it is carrying nobody. More idle vehicles spread through a service area make pickup faster, reducing the delay between requesting a ride and entering a car. Keeping those vehicles available, however, costs driver time and money.
Watch film
One ride has two different clocks. For the passenger, the trip begins with a request and includes waiting to be matched, waiting for the assigned car to arrive, and riding to the destination. For the driver, the service cycle begins while available, continues through pickup and transport, and ends only when the car is ready for another match.
Watch film
An on-demand car, a shared ride, and a fixed-route bus use vehicles very differently. An exclusive car offers direct travel but spends substantial time and distance on each passenger. Pooling shares some of that work at the cost of detours and coordination. A bus gains scale when enough people use a common corridor, but offers less flexible access.
Watch film