Your Next Car Will Most Assuredly NOT Drive Itself

In the February 2012 issue of Wired, Tom Vanderbilt authored an article titled “Let the Robot Drive: The Autonomous Car of the Future Is Here.”  The front cover of the magazine featured a robotic arm grasping a 5-speed shift knob with the surrounding type proclaiming in all caps, “YOUR NEXT CAR WILL DRIVE ITSELF.”  Mr. Vanderbilt followed up his headlining article three weeks later with another related one, posted to the Wired website and titled “Five Reasons the Robo-Car Haters Are Wrong.”

Not wanting to dive into this fray without preparing, I Googled Mr. Vanderbilt and was delighted to find that he had also authored the book Traffic: Why We Drive The Way We Do (And What It Says About Us).  Traffic had received a round of recommendations in the car mags when it was first published, but I had since forgotten about it.  Hungry for knowledge, I tracked the book down, bought a copy, and devoured it at my first opportunity.

Traffic has nothing to do with robotic vehicles, but very much to do with the dangers of leaving your home in a motorized metal cocoon and merging with the flow of crazies, nearly all of them distracted, unprofessional, and unfamiliar with their cars’ limits, but equally convinced they are every one of them “above average” in the performance of the duties necessary to drive from point A to point B.

I highly recommend the book—just reading it will make you a better and safer driver.  However, Mr. Vanderbilt is wrong to claim in his Wired article that the autonomous car is “here,” and Wired is wrong to leap to the conclusion that all of their readers’ next cars will be self-driving.

First, let me rattle off the “Five Reasons” I would be wrong, were I a “Hater.”  Vanderbilt presents these by listing each of the haters’ arguments, and then refuting each one.  Hater Point #1: Autonomous vehicles will be inherently more unsafe than humans.  #2: People like driving.  #3: The infrastructure for autonomous cars does not exist.  #4: Autonomous cars won’t survive the litigious environment in the United States.  #5: When you ride in an autonomous car, you ride with Big Brother.

My reasoning has nothing to do with any of these.  After reading Traffic, one knows that these five arguments are factually and statistically invalid.  #1: Over 90% of vehicle accidents are caused by human error.  Remove humans from the driving process, and the majority of crashes simply do not happen.  #2: People by and large, do not like driving.  If they did, they wouldn’t be staring at their smartphones and/or paying through the nose for distracting in-car technology to in an attempt to mentally remove themselves from the road-going experience.  #3: Infrastructure is not a problem.  Autonomous cars make use of the existing transportation network—that is what makes them so attractive as a transportation alternative.  #4: The “litigious environment” argument is an admittedly strong one.  As Vanderbilt concedes in his Wired article, the automakers’ legal departments are at this point actually forcing conservatism on their engineers—they do not want to deal with the legal backlash that would most certainly accompany the sudden unleashing of massive amounts of automation onto an uneducated user market.  But lawyers cannot stop the economic laws of supply and demand—they can, at best, only slow them down.  Autonomous cars will be a part of our lives, sooner or later.  #5: Given the amount of government paperwork involved with owning and driving a vehicle, plus the existence of GPS navigation systems, satellite radio, and the cell/smartphone in their pockets, drivers were in bed with Big Bro long before autonomous cars came along.

My argument is not against automation, but against the assertion that my next car will be autonomous and/or that the age of full automotive automation has indeed arrived.  I’m quite certain it will not and has not.  Why?  Two reasons: cost and user competence.

First off, at no point in either of his articles does Vanderbilt discuss what this high level of automation costs.  His examples of automated cars include the VW Junior, the Audi TT Pikes Peak, a BMW that drove itself on the autobahn from Munich to Ingolstadt (presumably so BMW could show Audi that they can do automation, too), and Google’s autonomous Prius.  He also brings up the Mercedes S-Class as an example of a current production car that is not autonomous, but features an extremely high level of automation.  Other than the S-Class of course (a steal starting at $91,850), I am unable to find any published information on what creating these autonomous cars cost the manufacturers, let alone what their sticker price would be if they were sold to the public.

The average sale price of a new car in the US is right around thirty thousand dollars at present.  Unless the autonomous car marketed for the masses was able to exist in that arena, it would not be a sales success in the numbers necessary to significantly ease metropolitan traffic or prevent human-error accidents.  Why not?  For a somewhat similar example, we turn to the Chevrolet Volt.

The Chevy Volt can theoretically achieve infinite miles per gallon, so it offers a huge advantage over a normal small car (as an autonomous car does).  Instead of saving time (as an autonomous car would), it saves money.  Here’s the Volt’s problem: it stickers at over $40k, instead of less than $20k as its platform-mate the Cruze, shackled with an oh-so-20th-century standard internal combustion engine and transmission, does.  The Cruze is selling well, whereas Volt production was recently suspended because dealers are having such trouble selling off existing inventory that they are avoiding purchasing Volts from the factory.

My point here is that despite offering a massive advantage over a similar car, as say an autonomous Google Prius would, the Volt is a sales failure because consumers are not willing to pay the large premium that the massive advantage demands.  I’m sure if autonomous cars were inexpensive to create, information about their cost would be easy to obtain.  If the average person could afford an autonomous car, we’d have heard about it by now.

The second reason my next car will not be autonomous is that people are bad enough drivers as it is, let alone after a computer has been doing the driving for them for the last six months.  Then, as they’re leaving the office after a long day at work (or worse, as they’re leaving the bar after a few cocktails on a rainy night) their flux capacitor won’t boot up correctly and they’ll get a “PC Load Letter” error message.  The car will revert to manual mode.  Desperate to get home, they take the wheel themselves.  They’ve never turned the steering wheel before, don’t know where the turn signal buttons/stalks are, and aren’t familiar with the feel and feedback from the gas and brake pedals.  Their seat doesn’t place them in an optimal eye position.  Their mirrors aren’t adjusted.  And it’s been six months since their brain fired the “driving” synapses.  Now, they are a greater hazard to everyone else on the road than they ever were before autonomous cars existed.  As much as I adore embracing new technology, I refuse to be that person—I don’t harbor any illusions about my ability to drive well in an unfamiliar car after a long layoff, and neither should anyone else.

Thus, until autonomous cars are a) cost-competitive and b) flawless, never requiring manual input, my next car (and yours as well, I hope) will NOT drive itself.