A Brief History of On-Demand Transport

The new breed of on-demand transport services exemplified by Uber and Lyft is a major departure from what came before, evidenced by the legal challenges worldwide. To understand how Uber and Lyft differ from traditional forms of on-demand transport services, it is necessary to look to the past.

In this second post in our science of on-demand transport series, we look at three important classes of traditional on-demand transport services: hackney carriages; dispatcher-based taxi services; and dial-a-ride services. This will be done via a new diagrammatic representation of these services, which will aid us in understanding their differences.

Let’s begin with hackney carriages, which have existed since the prehistory of on-demand transport (records date back to 1654). A key characteristic of hackney carriages is that passengers notify drivers that they desire transportation by hailing the driver on the side of the road. Regulation of the number of hackney carriages is performed via commissions such as the Public Carriage Office (PCO) in London, and the New York City Taxi and Limousine Commission.

1920_Hackney_CarriageIan Britton

Fig. 1: (Left) An original hackney carriage and (right) the modern version.

In a hackney carriage service, there are three actors (or agents) that interact: passengers; drivers; and the provider (in this case, the commission such as the PCO or an owner of the licence to operate a taxi). It is helpful to be explicit about what happens when a passenger is serviced by a hackney carriage:

  1. A passenger hails a driver, which notifies the driver that the passenger desires service.
  2. The driver decides whether or not to pick-up the passenger, which is followed by a negotiation between the driver and passenger on whether the price of the journey is acceptable.
  3. If the passenger accepts the price, the driver transports the passenger.
  4. On arrival, the passenger pays the driver.

Note that many steps in this hackney carriage protocol are information exchanges; e.g., when the passenger hails the driver, or in the price negotiation.

There is also an additional financial transaction: each driver pays part of their earnings to the provider at regular intervals. Importantly, this payment is not on a per journey basis; instead, it is typically longer, such as monthly.

Our discussion so far suggests that there are two key exchanges: financial; and information. Moreover, exchanges can be performed on a per journey basis or long-term basis (e.g., monthly). For instance, the financial exchange between passengers and drivers is on a per journey basis; while the financial exchanges between drivers and the provider are on a long-term basis.

These financial and information exchanges characterize the hackney carriage service. We can summarize these aspects diagrammatically, which leads to Fig. 2,  where we represent per journey exchanges by an arrow and do not label long-term or the absence of an exchange. Observe that in the hackney carriage services, there are only per journey exchanges between passengers and drivers, with none involving the provider. This is a unique feature of hackney carriages, as we will see in the dispatcher-based taxi and dial-a-ride services that we will consider next.

hackney_exchangeFig. 2: Exchange diagram for hackney carriage services.

Now we turn to a second class of traditional on-demand transport services: dispatcher-based taxis, which are a fixture in cities worldwide. In contrast to the hackney carriage, dispatcher-based taxis do not need to rely on passengers hailing from the side of the road. Instead, a passenger can call the dispatcher, who will notify drivers that a passenger desires a journey. This has the important advantage that passengers that are not in visual contact of drivers can still obtain a journey.

The protocol for passengers to obtain a ride is slightly different from the hackney carriage.

  1. A passenger calls the dispatcher requesting transportation.
  2. The dispatcher forwards the request to one or more drivers, and a decision is made as to which driver will transport the passenger.
  3. The selected driver travels to the location of the passenger, and negotiates with the passenger over the price.
  4. If the price is acceptable to the passenger, the driver transports the passenger to the requested location.
  5. At the end of the journey, the passenger pays the driver.

As for the hackney carriage service, there are two types of exchanges: financial and information. This suggests that we should again use an exchange diagram to characterize the service. These exchanges are illustrated for dispatcher-based taxi services in Fig. 3.

dispatcher_exchangeFig. 3: Exchange diagram for dispatcher-based taxi services.

Notice in Fig. 3 that we can immediately see that there are important differences between the dispatcher-based taxi services and the hackney carriage exchanges in Fig. 2. In particular, the provider is now involved in per journey exchanges. This occurs for two reasons: the provider (or dispatcher) is paid a commission for each passenger a driver transports; and the provider forwards each request by passengers to drivers.

Also observe that there are strong similarities with the hackney carriage services; in particular, there is still the negotiation between passengers and drivers, which is clear from the information exchange. Moreover, there is a direct payment from passengers to drivers. This allows us to clearly identify that dispatcher-based and hackney carriage services are both a form of taxi.

The final class of traditional on-demand transport service that we will survey are the dial-a-ride services. While less recognized compared with the taxi classes, dial-a-ride services play an important role throughout the world, particularly in aiding the elderly and disabled to travel, and also to fill gaps in supply in rural areas.

Key features of the dial-a-ride services are that they are not profit-driven and drivers are salaried. This means that there are no financial transactions involved on a per trip basis, which arises because the organizations that provide the service are often subsidized by municipalities or governments. This feature means that dial-a-ride services are fundamentally different from the hackney carriage and dispatcher-based taxi services we have considered so far.

The protocol for dial-ride services is as follows:

  1. A passenger requests a journey.
  2. The provider adds the passenger into the route of a driver.
  3. The provider notifies the driver of a change in their route.
  4. The driver collects the passenger at the appointed time, and transports the passenger to their desired destination.

As for the hackney carriage and the dispatcher-based model, we can represent financial and information exchanges present in the protocol diagrammatically, which reveals a number of important differences with the other traditional on-demand transport services.

dialaride_exchangeFig. 4: Exchange diagram for dial-a-ride services.

In particular, Fig. 4 shows that there are no per journey financial exchanges, which arises due to the fact that passengers do not pay for individual journeys. Another feature that sharply contrasts with taxi services is that there is no direct negotiation between passengers and drivers; instead, passengers make requests to the provider, and the provider then directs drivers without any negotiation. It is worth emphasizing that these fundamental differences between dial-a-ride services and taxi services lead to exchange diagrams that are visibly different (compare Fig. 2,3, and 4). This suggests that the exchange diagrams are a useful way to characterize on-demand transport services.

To conclude this post, let’s recap. We have surveyed three key traditional on-demand transport services; namely, the hackney carriage, dispatcher-based taxis, and dial-a-ride services. We observed that in each case, the service can be effectively characterized by considering financial and information exchanges, which occur on a per journey basis. We then represented these exchanges diagrammatically, which clearly illustrated the main differences between these services and forms a basis to classify different on-demand transport services.

In our next post, we will return to the new breed of on-demand transport services, exemplified by Uber. There, our exchange diagrams will provide a useful tool to understand more precisely just how different Uber is from the traditional approaches to on-demand transport.

Towards a science of Uber

Uber is the giant in the room for on-demand transport. Its rapid growth is challenging traditional approaches to on-demand transport. In particular, Uber, Lyft and local services such as Liftago in Prague are revealing that for many passengers and drivers, existing taxi services do not provide the features that they want.


One of the key features of Uber’s approach is dynamic pricing; that is, the price passengers pay and the amount drivers are paid varies depends on demand of passengers and supply of drivers—an example of a market-based approach. Although many details of the mechanism are not transparent, there is evidence that Uber’s approach is beneficial for some passengers and drivers in some regions, but, crucially, not all.

The fact that Uber’s approach is not universally superior is important. What it reveals is that there are many factors that determine the success of an on-demand transport service, and that this success is greatly dependent on the region it is implemented. This suggests that competitors to Uber, exploiting services based on different mechanisms, can overtake Uber in some regions.

In this series of posts, we introduce a methodology to compare and help recommend the right on-demand transport service to implement in a given region. We believe that our methodology will aid new providers and municipalities to decide on what classes of services are best suited to their region.

In the next post, we take the first step of identifying the different classes of on-demand transport services. This will form the basis for our comparisons in later posts.

Part II: A brief history of on-demand transport