Google's Driverless Car

Without much fanfare, Google's driverless car reached a significant milestone last summer: 700,000 miles without a single accident. That's 28 trips around the world — driving on the equator, if that were possible — and to the Moon and back and back to the Moon again, if that were possible (and given Google X, who knows …)

I've encountered Google's self-driving cars in the wild — they are hard not to see on Shoreline Blvd in Mountain View, Calif., where Google and LinkedIn are neighbors. And they are impressive. It's almost hard to believe it's not all an elaborate hoax as these vehicles coax their way through traffic without hitting or being thrown off by obstacles as diverse as red lights, cyclists and other vehicles. They can track hundreds of objects simultaneously — a feat no human can match.

Nobody really expects Google to get into the car business, but they do have proven method of insinuating themselves in industries which matter to them. Android, the mobile operating system, went from zero to world domination in a few short years. Microsoft may make more direct cash money from Android than Google, which gives the OS away. But a strategy of even leaving money on the table to be everywhere gives Google a tactical advantage that could pay huge dividends.

Samsung, HTC and other big handset makers lined up to power their phones with Android, to piggyback Google's self-interest. Now Google seems to be hoping for the same effect with the auto industry.

Reuters reports that Google is in talks with "most of the world's top automakers" in an effort to bring self-driving cars to market by 2020, and is putting together "a team of traditional and nontraditional suppliers" to make it so.

"We'd be remiss not to talk to ... the biggest auto manufacturers," Chris Urmson, director of Google's self-driving car project, told reporters paul Lienert and Joe White in an interview. "They've got a lot to offer." Which may be the understatement of the year. The car industry may not have the highest barrier to entry, but it is right up their with oil and aviation. It's not impossible, of course. But the jury is still even out on Tesla, a tech, Hollywood and momentum investor darling which Elon Musk says won't be profitable until — by sheer coincidence — 2020. Which is just another way of telling investors they are still backing a dream.

Some dreams come true, but I'm very skeptical about self-driving cars, for two main reasons:

• It's not monumentally better than human-driven cars equipped with better driver assistance systems such as lane-drifting warnings, drowsiness detection and adaptive cruise control, all of which are more easily deployed
• There are no half measures: To make a business of it, self-driving cars have to be street legal everywhere, which means overarching federal legislation or universal state and local acceptance

The first point significantly influences the second: big projects that upset everything, like a national highway system or dams, get done because the upside is huge: A big problem is solved or society benefits in a way that was unprecedented. Nobody knew they wanted a smartphone — there was no problem in the mobile phone space, per se. But the smartphone clearly addressed unarticulated desires, and, not insignificantly, killed the feature-phone business led by Nokia that was thriving with no end in sight.

What desires — articulated or not — do self-driving cars satisfy?

If driverless cars can reduce accidents, that would be a compelling argument. Consumers respond to safety ratings, especially if they have children. But rate of death and injury from car crashes has been in steady decline, even as the number of cars on the US roads steadily increases, and with very little emphasis on advanced driver assistance systems.

The greatest overall cause of car accidents arguably is driver skill and impairment. If there was no driver then no impairment — including the effects of drugs and alcohol — would contribute to accidents.

But can we get from here to there. Should we? Disruption is a tricky thing. Back when satellite radio was launched the competition was only terrestrial radio. Satellite was compelling because all programming was available everywhere — not constrained by the limited reach of broadcast signals — and it offered audio quality that rivaled FM.

Then came the perfect storm that was the mobile internet, Pandora and smartphone apps. As good an idea as satellite radio was — and it's still a nearly $4 billion business — I wonder who'd back such an idea now since simpler, cheaper equivalent solutions have emerged from the primordial tech goo.

Self-driving cars may be a good idea. But, like satellite radio it could also be that they won't be a good idea for too long. By 2020 cars may be much smarter in the right ways. Maybe in the next five years we'll all realize cars don't have to be smart enough to be given the keys entirely. And with no compelling reason to take what will still be an enormous, collective leap of faith, maybe we won't really want to.

There's another good reason to keep the driver in the driver's seat, as a commenter notes: Liability. If a self-driving car is in an accident, who — what — is to blame? I wouldn't want to be sued if I didn't play an active role. So who's the exposed party? Google? The automaker? The car?

The carmakers are smart to talk to Google, because they have tremendous street cred: It is determined, has resources, dreams big and sometimes puts progress before profits. But five years is a long time. My money is on the less sexy engineering required to make cars better co-pilots, protecting us from unpredictable dangers and ourselves with a grid of smart technology working together.

Google itself is also leading that charge with a version of Android that would be built directly into cars. As reported last month by Reuters, there is no timeline for that initiative. But it could complement or eclipse self-driving tech. It's a perfect hedge.

Still, I hate stamping out someone's dreams. Who am I deny anyone the unparalleled sensation of freedom that is being driven around.