If you are in the self-driving car business and you have not read Tom Vanderbilt’s book, Traffic - Why We Drive The Way We Do, then I am going to assert that you are not yet in the self-driving car business. One of my intellectual heroes, Tyler Cowen wrote a blurb on the book’s jacket that perfectly echoes my sentiments.

Everyone who drives—and many people who don’t—should read this book. It is a psychology book, a popular science book, and a how-to-save-your-life manual, all rolled into one. I found it gripping and fascinating from the very beginning to the very end.

Indeed, what may be most interesting about our car culture is how authentically interesting it is by any metric you can think of and yet how little attention and contemplation it gets. If you were out somewhere and someone pulled out a gun, stuck it in your face, and said, "I could blow your brains out right now", that would be one of the most interesting stories you’d tell people for the rest of your life, assuming you lived. Yet an exactly equivalent thing happens to me almost on a daily basis every time I get on my bicycle.

This book is extremely well written and researched. It is superbly packed with research findings that are absolutely fascinating, even shocking. Every page had some astonishing fact that just intensified my impression of the streets as a surreal nightmare. But a common theme emerged. This theme is best captured by the fact that in the New York Times review for this book, author Mary Roach (who is famous for snappy book titles) correctly points out that this book could easily have been titled simply "Idiots".

People are colossal idiots. I have always said that if you find someone who isn’t an idiot, you can change that by putting her or him behind the wheel of a car. My belief in this is so strong that I do not even except myself from this immutable law of human nature.

But there is more to the story. Much more. We’re not (always) idiots outside of our cars. The story, it seems, is not so simplistic and reductive. Why do cars make us idiots? What exactly is going on? Why have we built a transportation system that is so unflattering? How has this happened? The answer, as with most good plots, is that the situation is complicated. Damn complicated.

This point is pressed explicitly in an impressive spectrum of subtopics. The book points out "A better warning label for side-view mirrors: Objects in mirror are more complicated than they appear."

Another example is traffic reports. Traffic reports are interestingly related to my theory on economics. I have publicly said: "It may be that economics can be defined as: the philosophy of looking at the world in a way that if a reliable rule to predict the future is discovered, the rules instantly change." It turns out that traffic reports show the exact same behavior! What good are they then? (What good is economics?)

Traffic is engineered and as an engineer (an industrial engineer in fact), I appreciate the technical details. One hears about traffic flows and then flow rates and before long a hydraulic analogy starts to form. But this is wrong! Traffic flows are not like water flows at all. It turns out that traffic flows more like rice (yes rice) whose flow properties are even more mysterious than normal fluid dynamics.

A lot of times I have to hear about people who claim to like driving (I’ve yet to hear anyone enjoy the company of their fellow idiot drivers). This book conclusively debunked the idea that people generally view driving favorably. However, it is complex. Studies have shown that people say they dislike driving but also say they drive unnecessarily. How can that be reconciled?

The book covers a lot of ground. The serious complexities of the industrial engineering are almost trivial in light of the psychology of the human participants. I think the most important idea that I walked away with is that pretty much all traffic engineering is an exercise in programming human minds. This is why it is so challenging and mysterious.

The book was published in 2008 at the dawn of the autonomous vehicle zeitgeist, just a couple of years after the first DARPA Grand Challenge. It is a perfect summary of the complexities of transportation before the vaporous hype of autonomous cars really fogged up the view. Sure the book presciently interviews Sebastian Thrun and duly mentions autonomous vehicles, but at that time they just weren’t considered such a serious part of the equation. This is why autonomous vehicle engineers need to read this book, to understand where the real problems really are, independent of autonomous vehicles. Those problems are all human nature problems. It is my strong belief that the robotics technology has been ready for at least a decade. Dealing with people is another matter completely. I was not left optimistic that robots will ever integrate satisfactorily with the madness of human drivers. The book says much about technologies like ABS and ACC, "It’s probably no accident that whenever one hears of a smart technology, it refers to something that has been taken out of human control."

The book is filled with insights that do incidentally support autonomous vehicle enthusiasm. For example, rather than building more lanes, a more effective way to achieve higher traffic throughput is to have fewer crashes. This is standard traffic engineering philosophy, but I’m envisioning this being accomplished by removing the idiot from the process. This is a subtle but important point for autonomous technology.

This book really was just packed with all kinds of thought provoking details. Here are some miscellaneous notes I took which illustrate this.

  • Roundabouts are safer than Americans think. For autonomous vehicles? Interesting question.

  • Professional drivers have long term stress related health issues.

  • "Athlete parking" (what I call far off parking spots, the opposite of handicapped parking) is vindicated as usually better from a total time spent perspective. And you get some exercise!

  • Free parking is not free. And the whole Shoup Dogg’s can of worms.

  • Roads follow a "power law" where "a small minority of roads carry a huge majority of the traffic". I have noticed this and it is crucial to my belief that selectively modifying infrastructure to accommodate autonomous vehicles is worth exploring.

  • "Children at Play signs have not been shown to reduce speeds or accidents and most traffic departments will not put them up." Why do they exist then? The municipalities seem to put them up. "They may have even been put up after a child was hit or killed by a driver, in which case it would probably be more effective to erect a sign saying just that." This is exactly the same story as Share The Road signs which really should say Cyclists Killed Here.

  • Quite a bit about Hans Monderman who is famous for, "When you treat people like idiots, they’ll behave like that." Is that why people are idiots in their cars? It could be. I prefer to treat them as unconscious. To me they will always be idiots.

  • Though there’s some base rate issues, more people are killed in NYC legally crossing at cross walks than jaywalking.

  • Pick up trucks are the most deadly passenger automobiles. I have mixed feelings about that.

  • The safer cars get, the more risk drivers take. See Peltzman Effect. This is especially noteworthy where the consequences of risky driving behavior are transferred to vulnerable road users (VRU). Can autonomous vehicles finally stop that effect?

  • Clean energy cars might make traffic problems worse. It’s analogous to peak oil which (if you were paying attention to prognostications in the 1980s) should be merrily curing greenhouse emissions by now. But rather than wrestling over the last few drops of oil there’s literally still plenty of energy to burn. By allowing a lot more cars to exist for the same pollution footprint, electric cars won’t be doing traffic any favors. By dodging one obvious problem, it is conceivable that another is created.

  • I’m struck by what traffic engineers are optimizing for. Pedestrian safety? Vulnerable road users? Pollution? Wildlife? A pleasant place to live? I think they’re just being asked to move as many cars as possible. But is that really what we should be aiming for?

While I really liked this book and recommend it to everyone, there were some things that I would have liked to see. The author sometimes describes charts and plots. He’s a good author, but why not just include the plots?

I also thought it was odd that he never even mentions what for me is the real elephant in the room—how much life is lost to driving itself? In a normal career which includes a normal commute you will spend the equivalent of about 600 complete days worth of waking time imprisoned in your car. In your car commuting. That seems pretty serious to me, especially when almost everyone is afflicted with the disease and the cure is so attainable!

The book covers some of the inputs that people use to form their own calculation of risk (Ever been hit before? Know anyone who died in a crash? Etc.) One factor not mentioned that must surely be huge is that consumers are continuously pumped full of advertising propaganda by car makers distorting the true safety situation.

Still, this book is a great introduction to the extreme complexity of our transportation culture and psychology. To paraphrase a line quoted in the book, traffic is not rocket science—it’s more complicated.