Tuesday, 5 November 2013


I was reading a review of a new biography of Charles Manson the other day and was struck by how often the direction of his life turned on stolen cars and police check-points. Though we associate the growth of car-ownership and the highway system in postwar America with freedom, the reality was increased state surveillance and control, which has an obvious analogue with the Internet of today. Daniel Albert captures this well in an article on the coming robot chauffeur: "A century ago, as cars suddenly made it possible for Americans to slip local bonds, government responded with rules and regulations to keep track of who was driving what. Whether intentionally or not, traffic regulations became a national dragnet serving all manner of police purposes, suppressing everything from sex trafficking to adolescent rebellion".

Albert is a fine satirist as well as a car-geek (his take on flying cars is a hoot), but he could still learn a thing or two from Allister Heath, the editor of City AM, who told us last week to "Forget high speed rail. Driverless cars will revolutionise transport". The chief boon of this revolution is that "Driverless cars will reduce accidents by 90 per cent by eliminating human error". As a free-market evangelist, Heath is shrewd enough to recognise a corporatist gravy-train like HS2 when he sees one, but he dons rosé specs when arguing the case for autonomous vehicles, particularly when you realise that the government-backed trial planned for Milton Keynes (which triggered his panegyric) involves vehicles that are little more than golf-buggies (you could argue that a top-speed of 12mph will inevitably reduce accidents, but that's not what he is suggesting).

Though he doesn't provide a link, the 90% claim can be traced back to the Eno Center for Transportation, a US lobby group that bills itself as "a neutral, non-partisan think-tank" (i.e. it's funded by transport and technology companies). The figure originates not in a peer-reviewed study but in "a competitive paper competition among Eno’s Leadership Development Conference Fellows". Presumably the competition was to see who could be most on-message. The central claim in the paper, in respect of US fatal car accidents in 2011, is that: "Over 40 percent of these fatal crashes involve alcohol, distraction, drug involvement and/or fatigue. Self-driven vehicles would not fall prey to human failings, suggesting the potential for at least a 40 percent fatal crash-rate reduction, assuming automated malfunctions are minimal and everything else remains constant".

Simple logic would suggest that "at least" should be "at most", while the assumptions are asking a lot. The 40% figure is opaquely derived from a table of multiple causes - i.e. a crash may be the result of both drink and drugs. Alcohol is a factor in 31% of fatal crashes, distraction a factor in 21%, drugs a factor in 7% and fatigue a factor in 2.5%. You could make the figure of 40% by adding together alcohol, drugs and fatigue (perhaps the real benefit of driverless cars will be drunken snoozing), but it looks more like a guesstimate. It also implies, given that alcohol is the chief factor, that integrated breathalysers could deliver most of the safety benefits promised by robot chauffeurs.

The paper further observes that "Driver error is believed to be the main reason behind over 90 percent of all crashes". This is the opinion that has morphed into Heath's factoid. The objection to this "belief" is that it ignores who actually dies or is injured in car crashes. According to the US government, there were 29,757 crashes resulting in 32,367 fatalities in 2011. The Eno Center paper misrepresents the latter as "Total Fatal & Inurious [sic] Crashes per Year in U.S." (not only wasn't it peer-reviewed, the paper may not have been proof-read). Of the deaths, 22,448 were car occupants, 4,612 were motorcyclists, 4,432 were pedestrians and 677 were pedal-cyclists, with 198 other/unknown. Given that 30% of all fatalities were non-car-occupants (and this figure is probably similar for non-fatal injuries), who will continue to be the cause of many accidents and will presumably not all be augmented by technology, it is a stretch to believe that driverless cars will save two-thirds of them from their own folly. A robot may be able to brake quicker than a human, but stopping distances will be no shorter. If you step straight in front of a robot car, you'll still get hit.

Heath then proceeds to roll out a series of popular nonsense claims in respect of driverless cars: "They will free up commute time for work, rest, sleep or entertainment, unleashing a huge increase in productivity and revolutionising the economics of commuting and thus of cities". This is the Winnebago claim, the idea that cars will evolve into a combination of office, gym, kitchen and bedroom on wheels. Obviously the editor of City AM has not thought through the economics of this, given the inescapable fact that a Winnebago costs a lot more than a Ford Fiesta. Or perhaps he has ...

He goes on: "Many workers will be able to live much further out and even commute for hours. Driverless cars will be able to drop passengers off and then wait for them at a convenient location, returning to pick them up; it may even be that they will be rented, rather than owned, which could do away with the need for homeowners to worry about parking spaces". This is the corral claim. Massive parking lots in the suburbs, and at the edges of city centres, would increase traffic miles and fuel consumption far in excess of the gains from more efficient robot driving. In practice, most city commuters will continue to use mass transit systems, i.e. buses and trains, which will remain the most efficient use of resources per capita. While these may have robot drivers in future, they won't have beds or desks, so "commuting for hours" won't be any more attractive than it is today.

It seems odd that a Hayekian libertarian like Heath should be a proponent of driverless cars, given that their credible benefits (in terms of safety, crime and road capacity) derive from a reduction in personal autonomy. Even the idea of "eliminating human error" has a distinctly un-Hayekian air to it. A robot chauffeur is still a chauffeur, so perhaps the attraction is the idea of command, with the hint that this will be an elite preference. Or perhaps this is just the automophilia that drove Margaret Thatcher's distaste for public transport. In a Telegraph article in April, Heath espied the future: "In 20 years’ time, the demand will be for far more car journeys and more roads, reversing the trend of the past decade and the recent resurgence of the railways. We will be entering a new golden age for car travel". In 20 years, Charles Manson will be 99. By then, car-thefts and police road-blocks may have been consigned to history.

No comments:

Post a Comment