Yesterday on NPR, there was a story about how the National Highway Traffic Safety Administration had decided the computer system that drives a “self-driving car” can be considered a “driver” for regulatory purposes. In the segment, several drivers were interviewed, and all but one hated the idea of a self-driving car. They didn’t want to surrender “control.”
Never mind that control is an illusion, one that causes traffic accidents and gets you killed. Never mind that when things happen too fast or beyond your ability to comprehend, you willingly, gladly surrender control on a daily basis. Some studies estimate that 90 percent of fatalities could be prevented by self-driving cars.
This future assumes a certain degree of mass adoption. If there's still one jackass on the road with a New Jersey license plate who drives like a maniac in my genteel town, confusing the locals and the self-driving cars alike, someone is likely to get killed. But let’s ignore safety for now because no one really thinks about it until it's too late. (That's why everyone buys security products instead of observing good software development practices.)
No, my dear readers, let’s talk about something that means more to all of us than dear life itself: Traffic.
Consider a world in which everyone uses self-driving cars -- the utopia imagined by such intellectual giants as the people behind "Hot Tub Time Machine 2." Now consider maybe the second evolution of the self-driving car, the fully connected self-driving car. Sure, your mind will start to wander to some sci-fi pic you saw or the last press report about “hackers ahhhh!” instead of “barely evolved monkey brains using their instinct to control high-speed transportation devices ahhhh!”
I need you to focus and think about how these Ethernet packets are reaching you right now. Lots of people (better be lots or my editor will make me write another top 10 list) are receiving this transmission and many others across a vast network of tubes. Those tubes manage to operate much more efficiently and effectively than our highway system despite the many billions of dollars more spent on the latter. You barely have any control whatsoever over these packets or the packets you sent requesting these packets. That is why it works so much better than our highway system.
In fact, when you surrender control, you allow an algorithm used by such sophisticated creatures as ants to optimize your packets. Yes, the same algorithm used by ants to control how many ants leave the hive is similar to that used to govern the “quality of service” on your network devices. Basically, when each ant comes back into the hive, it touches another ant, which signals that it's OK to send another ant out of the hive. Packet control works in a similar way.
In a world where interconnected self-driving cars have reached mass adoption, our cars will be packets and the road an Ethernet cable. Packet control and proper coordination of the network will prevent traffic congestion up to the capacity of the network. Latency will be injected at the right moments. Basically, the system will add space between you and the car in front of you and adjust to an appropriate speed for that particular situation -- which in some cases may be north of the current 70mph limit. In fact, there will probably be increasingly fast road networks with faster self-driving cars.
Almost as important as traffic, with near-perfect data, you can be virtually guaranteed an on-time arrival based on a particular departure time. And you millennials can text in peace without an annoying steering wheel in your way! With near-perfect data, capacity planning for the highways and local roads will occur using more or less linear algebra.
Thanks to the centrality of the automobile in American life, this model could expand beyond roads and into the other areas of civic planning. Knowing how many people are coming into town would allow hotels to quickly adjust prices and compete with AirBnb. It will allow stores to preemptively alter staffing.
While the average driver is reluctant (the way the public was leery of cellphones, microwave ovens, and streaming video at first), it seems to me as if everyone in tech thinks self-driving cars are coming in the next five years or so.
What I find truly ironic is that those same tech people often resist the cloud, ignore new data and machine learning technologies, or balk at automating deployments. Plus, the businesses they work for still lovingly craft analytics in spreadsheets and drive their sales processes on “gut” instinct. Why? Because they want control.
To me the question isn’t whether we’ll be better off, but whether the self-driving car reaches mass adoption before the bulk of businesses modernize their sales, marketing, and internal operations to be data-driven. The self-driving car isn’t ready yet, but we have the tools and techniques to “self-drive” most companies. So join me -- and give up the illusion of control.