Saturday, January 24, 2009


In computer networking, there's a heuristic known as Metcalfe's Law, which says that the value of a network is proportional not to the number of users, but to the number of connections between users, which is proportional to the square of the number of users. Each user gets some benefit from being able to communicate with each of the other users. Now that's not quite accurate when you look at the details, since not every pair of users is equally likely to want to talk, but it's still a super-linear function of some sort.

The same thing can be said of transportation networks. What matters is not the number of stations but the number of origin-destination pairs connected, or the number of potential trips served. In mass transit, it's impossible to take everyone directly where they want to go, since presumably different people want to go different places, and you can't give everyone their own vehicle. So the key to a successful mass transit system is providing good connections. Connections inherently require walking, waiting, and there's always a risk of missing your connection, so for an effective connection, they have to be minimized.

Switzerland is one place that gets connections very right. The entire rail network runs to a single hourly timetable, which is carefully designed to maximize the number of timed transfers between different services, including local and express trains on the main rail network, as well as a multitude of minor lines. Everything is absolutely coordinated, even across different rail operators, and connections are not only timed but in many cases guaranteed, with trains waiting for each other when possible.

The same applies to high speed rail systems all over the world. They all have either extensive through running from the high speed lines onto the conventional network, or else an extensive system of connecting rail lines. France's TGV network is extensive, but much of that extent lies outside the high speed lines themselves. The TGV trains run on the conventional network for part of their journey, and use the LGV as a shortcut to significantly reduce trip time on the final leg to Paris. Even in Spain, there is through running between the AVE high speed network and the conventional lines, with special gauge-changing trains to overcome the difference in track gauge. In Japan, there is only limited through running due to the incompatible track gauge and loading gauge, but the conventional rail network has very extensive and frequent service to provide connections.

Even the rather small intercity rail network in California is augmented by extensive connecting service (with through ticketing). This service is mostly provided by Amtrak buses, but in some cases also by local transit services, and even via the ACE commuter rail. Of course, much more can be done here, especially in terms of integrating Metrolink, Caltrain, and Coaster into the intercity rail system.

But what of High Speed Rail? From my understanding of the proposal, there is no thought at all given to through-running to non-HSR network, which is unfortunate, as it means that passengers travelling, for example, from Sacramento to Los Angeles will be forced to transfer until the line is extended to Sacramento. And those in Redding or Palm Springs are never going to get direct HSR service. There's also very little in the planning documents about how HSR will interact with the existing intercity rail system. Of course, it would help if the patchwork of existing rail services could be better integrated too, which is a matter for a future post.


  1. Arcady

    Very good explanation of timed-transfer, which I've been doing as a bus network planner for 20 years. See my complementary post here:

    Cheers, Jarrett

  2. I totally agree that the lack of high-speed tracks to somewhere should not preclude the high-speed trains running. The bigegst cost would be electrifying the tracks, and the potential ridership would have to be enough to justify that cost.
    Transport is all about connecting people to places. Everything else is secondary.