Tom Vanderbilt has been talking about traffic for several years now and he highlighted again in a recent talk what leads to traffic and congestion:

Tom Vanderbilt, author of Traffic, gave a great 20-minute overview on the counterintuitive science of congestion at the Boing Boing: Ingenuity conference in San Francisco last month. Turns out a lot of the problems we ascribe to poor roads or other drivers are really our own fault. “[T]he individual driver cannot often understand the larger traffic system,” says Vanderbilt…

In fact, says Vanderbilt, traffic would be much better off if cars stayed in both lanes then merged at the very end, one by one, like a zipper. It’s safer (fewer lane changes), it reduces back-ups (often up to 40 percent), and it quenches road rage (still on the rise)…

A big reason for traffic is that too many cars are trying to occupy too little space on the road. But that’s not the only problem. A human inability to maintain a steady speed and following distance on the highway makes traffic a lot less smooth than it could be…

“You’re not driving into a traffic jam,” says Vanderbilt. “A traffic jam is basically driving into you.” He thinks autonomous cars will reduce this problem considerably...

That’s too bad, he says, because even a small drop in driving would improve congestion dramatically. One recent study of metropolitan Boston found that getting 1 percent of commuters off the road would enable the rest to get home 18 percent faster. Vanderbilt ends his science of traffic talk without suggesting ways to target this 1 percent. Fortunately there’s also a science of mass transit on the case.

In other words, individual drivers put their self-interests over the health of the entire system. So, then isn’t the trick getting drivers to recognize the larger system issues? Imagine signs at zipper merges where drivers were told to use all of the lanes – or even if this was the law. Or, if cities cut parking supply even further – this might prompt people to use mass transit more. Or, perhaps autonomous cars can really provide some solutions.

Another thought: this explanation of traffic sounds suspiciously like a sociological approach. The system is what is important when analyzing traffic, not starting with the individual drivers who will generally act in their own self-interest. New Urbanists tend to make a similar argument: roads should be designed less for cars alone, putting their interests first, and instead should make room for others like pedestrians, cyclists, and those living along the street.