For the past week, the entire San Francisco Bay Area has been on the brink of a full-fledged train strike. The unpredictable negotiations have left commuters in limbo: it’s too risky to schedule business meetings, but there is not enough uncertainty to arrange alternative transportation. As a result, BART has lost an estimated $168,000 after a 5 percent fear-based drop in traffic.
During last summer’s BART strike, a few outspoken Silicon Valley technologists incurred the wrath of the civil libertarian press for suggesting that San Francisco should simply automate the train operators and be done with the labor mess. Rather than join the techies vs. worker rage fest, I decided to simply ask the Bay Area Rapid Transit (BART) authority if it was technically possible to automate train drivers, just like Paris and other cities have done around the world.
BART’s answer was conspicuously dismissive. After exchanging emails for a month, they declined my request for an interview and instead gave me this statement:
“BART’s response to your inquiry is that you have posed valid theoretical questions best answered by researchers who study the future of train technology, not by the people who are focused on using the existing technology safely.
Thanks for your continued patience.”
In other parts of the world, unions and tech-happy politicians are unafraid to engage in a vigorous public debate over the cost and safety of automating trains. The mayor of London has proposed the tube go completely driverless by 2020, using Paris as a model of where unions can work out their opposition to the job-wrecking technology.
Our own Congress has proposed fully automated trains for the U.S. by 2015. But in San Francisco, the home of the robotic car, the silence is deafening.
So I decided to take BART up on its suggestion and ask around. I was surprised at how many people feared talking about the subject, but most of the experts who would speak said that it is definitely possible to automate BART and might be cheaper and safer, too.
“It would be possible to replace the train operators with automated control systems,” says Tom Rubin, a veteran transportation consultant in Oakland and former financial auditor of the BART system.
“There may be labor union issues that require personnel on the train,” said Peter Torrellas, Chief Technology Officer of Mobility and Logistics at Siemens. But “it is technically possible to have driverless trains that are safe.” Torrellas was forbidden by his communications team from commenting on the BART system. However, I got around their constraints by asking if there were a train system in the U.S. that couldn’t be automated. Torrellas said there was no technical barrier to automation.
Okay, there’s probably no technical barrier to automation, but is it worth the money? That’s hard to estimate, because the only way to really tell is if BART had the political willpower to commission a public report. Moreover, city projects are plagued with cost overrun, often by two to three times the initial estimates.
In Scotland, driverless trains were part of a $70 million total upgrade plan. Rubin was willing to give us a (very) cautious ballpark estimate, estimating that the government could save around $250 million.
BART will shell out $400 million in labor costs for the $1.6 billion transit system, which pays train operators some of the highest wages in the country ($66,000 – $74,000 a year). More specifically, there are roughly 500 unionized train operators and station attendants, averaging $92,156 a year, with benefits. Automation never fully replaces every worker, but BART would save a maximum of around $46,078,000 per year in labor costs — or more, if it ends up increasing the number of trains from 669 to 1,000.
Though Rubin says automation could cost tens of millions of dollars, ultimately it would save the city a lot of money. Siemens (which has a vested interest in automation) also concurred that automation saves cities money. Lastly, a spokesman from the International Association of Public Transport noted that metro automation typically saves 15 percent, but he couldn’t comment on BART.
We should emphasize that these numbers are ballpark and unverified since, again, BART has apparently never bothered to ask. BART may have some unknown and unique challenges that could make it much more (or less) expensive.
Aside from costs, safety is another serious concern. Even though BART is largely automated, train operators still need to check that passengers have safely made it through the doors and there are no track obstructions. Torrelles says that advances in object detection prevent automated systems from driving until all limbs are inside the train and there’s nothing lying on the track.
Not all experts I spoke to thought automation was advisable. Former Canadian IT manager for SkyTrain, George Slade, told me that there were managerial and physical issues at stake.
“With full automation most employees are really just waiting around most of the time waiting for something to break,” writes Slade to me through a Help A Reporter Out (HARO) inquiry. “This includes a lot of miss behavior [sic]. One of the other managers informed me when I started that working at SkyTrain was like being in high school again.”
Additionally, Slade argues that San Francisco has “secure and clear guide-way” on the track for the train. Rubin replies to us that securing the tracks would be a “(comparatively) ‘easy’ fix” and still be worth the roughly $250 million in savings over a decade.
The disappointing answer is that no one seems to know — exactly — if automating BART is worth the cost. One reason might be that officials are too afraid of union interests to explore the question.
Based on the responses I received, BART is either concealing facts about the possibility of automation, or its management hasn’t seriously considered it, which means BART is either negligent or ignorant. Both possibilities should concern tax payers.