The End of Rail

This year there were more large increases in the cost of rail tickets.

I was largely uninterested, though by no means disinterested, in the subject. Passenger rail travel is just inherently enormously expensive. The government taxes road travel heavily through road fund duty and fuel taxes, and at the same time subsidises rail travel heavily, and the net result is that for most journeys, rail is not an option, and for most of those that rail travel is available, it is about three times the cost of road travel. If you take the true cost, without the effect of the taxes and subsidies, rail transport is probably about ten times as expensive as road transport.

Privatisation was an obvious response to this long-standing situation, but the improvements were small, and the cost differential has only increased over time.

I've never clearly understood where the huge costs of rail come from. There are several sources I can think of, but whether some of them are insignificant, or if one of them dominates all the others, I'm not sure.

One obvious contributor is the safety standard that rail is held to. Rail travel is about a hundred times safer than road travel, and that is achieved through massive expenditure on signalling, inspections, pre-emptive maintenance and staff to supervise.

A second factor is evolution. Cars are sold in their millions, and the industry is subject to constant, intensive competition and improvements. Trains, by comparison, are rarely-produced items. The result is that trains evolve relatively slowly: a ten-year-old train and a thirty-year-old train are barely distinguishable, while cars of similar vintages are very obviously distinct. The manufacturing process is affected as well as the finished product.

The third is the lack of flexibility which defines rail travel. A road vehicle can go anywhere provided the ground is hard and flat. A train can only go where the network has been built for it. We do talk about a “road network” in the same sense as a rail network, but the concepts are not equivalent for that reason. A motorway might be planned and engineered to just the same degree as a railway line1, but the defining feature of road travel is not the motorway, but the driveway — the thing that allows anyone to join the road network at almost any point. That makes the road system an open network, while the rail system is a closed network, and the differences between road and rail are mostly the differences between open and closed systems. (That also covers the evolution factor above: vehicles for the rail network are selected by the network operator; vehicles for the road network are selected by the network users).

If road travel is so far superior to rail travel, why am I paying £4000 a year to sit on the train and write this?  One reason is of course the subsidies — but the commuter routes into London aren't directly subsidised; they run at a profit (though they gain from the subsidised local routes that feed into them). The fact remains that  for the one case of bringing a very large number of people to the same place at the same time, rail still has benefits. It has the advantages of a closed system as well as the disadvantages.

I am on a packed 12-coach train currently at St Pancras station. About a thousand people will get off within the next mile. Eight minutes ahead is another train, and eight minutes behind, yet another.

The chaotic open system of road cannot achieve the same peak efficiency in terms of use of space and labour, just as TCP/IP cannot achieve the same peak efficiency as a circuit-switched connection. That peak efficiency gap is very visible, whereas the very large cost gap the other way at lower levels of utilisation is not so visible. These packed carriages will spend most of the rest of the day empty or near empty, trundling around the network with almost the same cost in fuel and supervision. An unused car has significant storage costs, but because the costs are concentrated on the vehicle rather than the network, the overall cost is much more sensitive to levels of usage.

The other reason for taking a train rather than a car is that I would have to actually drive a car, whereas I can sit in a train and blog.

The crucial fact for the future is that the two decisive advantages that rail has, in a limited set of cases, over road — better peak space-efficiency and better labour-efficiency in the form of driving — are both on the decline because of the technology being applied to road. Self-driving cars are twenty years off at most, and I would expect nearer ten. That allows vehicles to travel faster and closer together, since human reaction times are taken out of the process of maintaining separation. I will be able to read, blog or watch television in a car as I do on the train. Parking costs will reduce because the vehicles will be able to disperse themselves until needed, instead of all having to compete for scarce parking space at the highest-density destinations.

That’s another way in which the rail/road comparison resembles the closed/open network comparison. Rail’s advantage is that it takes away the need for intelligence in the vehicles; road is best placed to take advantage of putting more intelligence in the vehicles. As intelligence becomes cheap, road could even exceed rail’s peak efficiency, by being more adaptive and responsive to conditions — using alternative routes or schedules. Rail’s throughput is limited by the flow of people through the chokepoints at stations more than by the capacity of the rails themselves.

Those changes take away any reason for passenger rail to continue to exist. (It’s possible that rail vehicles are more suited to very heavy loads, and so will remain useful for rail freight. I doubt it, but I don’t know enough to be sure).

Any forward-looking, integrated transport policy today will be oriented towards phasing out rail and preparing for self-driving cars.

Whether the Luton Guided Busway meets those criteria is anyone’s guess. It’s not out of the question.


 1. It probably isn’t, but let’s pretend it is.