In his book Faster: The Acceleration of Just About Everything, James Gleick takes readers inside
Airlines. There he shows how the hub-and-spoke model of air travel has given
way to point-to-point flights in an effort to maximize use of an airline’s fleet.
A single plane will criss-cross the country and rarely repeat any single day’s travel
log. The schedule Gleick shows readers is one not devised by human traffic
controllers but by a computer that is constantly, like modern GPS devices,
“recalculating.” Taking in up-to-the-minute information of travel delays, flight
crew schedules, cargo weight, winds aloft, and other data on all flights in the
fleet, the computer is constantly optimizing all the variables to keep planes
in the air instead of on the ground. System Operations Control
And computers like this one have largely succeeded in increasing efficiency. However, in the process, as Gleick shows, the margins for error are reduced to as small as possible. Because of the random criss-cross pattern of flights, a whole airline’s fleet is more and more interdependent. And with smaller and smaller margins of error, that interconnectedness actually shrinks the fabric of the fleet—like a new cotton t-shirt thrown in the dryer. These two dynamics—interdependence and smaller margins—create a new challenge for airlines: cascades. Gleick explains how a single delay in a regional flight can reverberate for days, causing delays around the country.
Gleick’s book came out in 1999 and illustrated the challenges of an interconnected technological system. For passengers, many of the delays they experience are nearly inexplicable because the system is so complex. And even if the reasons are understood, they are rarely explained to the passengers waiting at the gate or on the tarmac.
Now, nearly 15 years later, we are looking down the road toward a new technological system on the horizon: driverless cars. For the sheer novelty of it, many are enthusiastic about it. Imagining a panacea of free-time and new efficiencies, many are evangelizing for Google’s ambitious initiative. Already a number of
states including California
have passed legislation allowing for Google to continue testing on
heavily-trafficked roads. At last report, driverless cars had logged 300,000
miles. The latest news is that testing will begin in the UK by the end
of the year.
But just like that computer at American Airlines, a question remains: Who will coordinate the trips of the driverless cars of the future? The car’s own internal computers will not function on their own. For them to work at all, they will need to be connected to satellites, drawing down information about their location in conjunction with a digital map of every road available. In other words, every driverless car will have to be connected to the Internet in order to function properly. (Location information for all these cars will raise its own privacy questions, but we’ll leave that for another day.) Getting “out of range” and “losing reception” like a cell phone will no longer be an option for driverless cars. “Can you hear me now?”
Once driverless cars are the norm, they will need to be interconnected. So who will be coordinating their movements? It won’t be a human traffic controller. It will be a computer. Let’s call our computer “HAL.” For such cars to “play nicely together,” HAL will need to talk to every other computer that’s directing traffic. If Google is traffic-control headquarters, will there be room for competitors? Will HAL willingly work with another HAL? “I can’t do that, Dave.”
By contrast, GPS devices had numerous companies competing in the market, from Garmin to TomTom to Magellan to factory-installed devices. They didn’t need to communicate with one another. But driverless cars will have to: HAL will be essential. Will there be room for competing companies to provide this service? The needs of the technology itself seems to preclude having more than one HAL. Having a single provider will be the most efficient option.
If integrated communication wasn’t reason enough for having a single HAL, there’s another reason: optimizing the traffic itself. Traffic congestion won’t solve itself. Driverless cars, like airline fleets, will need HAL to be constantly recalculating and rerouting traffic. And we humans will want HAL to do this too. We don’t want to spend any more time in our cars than we have to. We will want HAL to run things efficiently. In exchange for efficiency, we may be willing to surrender that power. Even if that means that there’s only one HAL, and he’s knows everything, and has all the power.
But with HAL at the helm, will we find ourselves in similar situations to Gleick’s airlines—where the interconnected system is so efficient that it has very little margin for error? Will the t-shirt be so tight that we are no longer comfortable? And if the shirt is too tight, will marginal errors send ripples reverberating through traffic for days after? With HAL at the helm, will passengers find themselves waiting in traffic on a delayed schedule without any explanation?
These are legitimate questions we will have to navigate as driverless cars come into view. If we don’t, it will be HAL who decides.
What’s your technology deciding for you?
Technology’s Two Futures
Technology’s Two Futures