Cars will drive themselves before freight trains do

Self-driving cars are coming. Self- driving trucks, too. OK, maybe they're not coming tomorrow, but there certainly are a lot of resources and smarts being devoted to separating us from our steering wheels.

What about trains? They travel on tracks, which you might think would make automating them easier. Some urban-rail and high-speed train lines already are partially or even entirely automated -- and of course on the Island of Sodor the whole system is, sort of. The U.S. railroad industry is also spending billions of dollars to install a Congressionally mandated automatic braking system to prevent accidents.

So self-driving freight locomotives are right around the corner, right? Uh, no. Lance Fritz, the new chief executive officer of the biggest U.S. freight railroad, Union Pacific, was at Bloomberg HQ Wednesday morning. (Here's a report on the visit from Bloomberg's Thomas Black.) When I asked him excitedly about driverless trains, he replied, "We're not in the process of developing that or trying to develop that." Not that it wouldn't be cool, he allowed. "There might be an economic justification," he said. "There might also be a safety justification." But no, it isn't a priority.

Why not? Science journalist Carl Franzen got to this question before me in a well-researched article for Vice's Motherboard in January, so I'll start by reciting his reasons: -- Those automated and semi-automated urban rail systems and high-speed trains run on limited networks that are cordoned off from troublemaking cars and humans. Railroads that use vast open rail networks have to worry a lot more about what's on their tracks and whether the tracks are damaged.

-- Moving to driverless cars would free billions of man-hours a year for other purposes. Train engineers are few enough in number that you wouldn't see similar economic gains.

-- Those train engineers tend to belong to unions that don't look kindly upon innovations that eliminate jobs.

-- People are weirded out by the idea of driverless trains.

Those explanations all accord with what Fritz said Wednesday. But there was another reason that emerged from his remarks that seemed really important. It's that, unlike the developers of driverless cars, Union Pacific controls both the trains and the network upon which they travel. And its CEO seems to think that improvements in the network (Union Pacific's consists of 32,000 route miles) will bring bigger safety and productivity gains than increased automation of the trains.

Consider the Congressional mandate to implement positive train control to keep trains from running into each other or speeding. It was a reaction to the deadly Chatsworth train collision of 2008, in which a Metrolink commuter train and a Union Pacific freight train plowed into each other in the northwestern corner of Los Angeles (federal investigators concluded that the Metrolink engineer had run through a red signal while texting). PTC is meant to automatically apply a train's brakes if a collision with another train is imminent, if it is traveling too fast or if it is headed into a work zone.

PTC might have prevented the Amtrak crash in Philadelphia on May 13, but it is still months away from implementation along Amtrak's Northeast corridor. The freight railroads' even-more- involved PTC system (Fritz called Amtrak's version "PTC-light") is still a couple of years from completion. Union Pacific has already spent about $1.7 billion on it, with an estimated $800 million to go.

Fritz said he is hopeful that PTC will work as planned in preventing accidents in a way that "doesn't bring the railroad to its knees." But, he added, "there are ways you could spend $2.5 billion that will save more lives, specifically at grade crossings," he said. PTC will do nothing to prevent trains from plowing into cars at railroad crossings; making it harder for cars to cross when a train is coming would.

Another big issue is derailments because of track flaws, which most often occur where sections of rail are welded together. "Rail health is really important to us," Fritz said. For decades, railroads have bought 80-foot sections of rail and welded them together. Starting this spring, Union Pacific is buying 480-foot sections from Nippon Steel that are transported from Japan to Stockton, California, in a special ship. Which will eventually mean about one-sixth as many welds on Union Pacific's rail network.

That's what a company that controls its own transportation network chooses to spend money on. One result is that the U.S. freight rail infrastructure gets much higher grades than its roads or airports. Another, though, is that rail executives will make different, less-speculative investment decisions than those at companies that want to break into or transform the auto market, where the network is (mostly) public property. Google's huge investment in mapping has no real parallel in the railroad industry, for example. "We map our railroad," Fritz said, "but not GPS to-the-foot mapping." This seems like a rational choice. But it and choices like it mean that railroads are less likely to see the kind of transformative changes that may be in store for cars during the next few years. Then again, it may take decades for those transformative changes to play out, and they could be a disappointment. It's sustaining innovation disruptive innovation. Disruptive innovations are where the real excitement is. They are also a lot harder to plan around.

Justin Fox is a Bloomberg View columnist writing about business.