American Scientist

To the Editors:

I always enjoy Brian Hayes for his fun-filled mathematical exercises, and "Playing in Traffic" (Computing Science, July–August 2015) is no exception. As he shows, one can’t always plan a route using traffic reports, because traffic is frequently bad on all roads. Indeed, building more roads and bridges doesn’t seem to ease congestion, because there are times when all routes tend to fill up. Yet we still build, and now the New Jersey Turnpike has 14 lanes in one highly traveled section. High-density traffic occurs at peak times: morning and afternoon rush hours, beginning and ending weekends, and evening and weekend shopping. Often, the only choice is when to go, not how. Nonurban areas will have limited roads and choices, so traffic may be just as bad.

The safety rule of leaving one car length between you and the car in front for every 10 miles per hour of speed is largely ignored. On a limited access road, with a speed limit of 60 miles an hour, a safe maximum density is 14.3 percent (one car for every seven spaces). However, “selfish” driving guarantees that someone will fill what seems to be extra space. So seven spaces shrink to three and a half, effectively doubling the capacity to 28.6 percent. Traffic still moves reasonably well, but with decreased safety, increased tie-ups, and slowing traffic at exits and entries. Pileups of 50 to 100 cars are not that uncommon, and in the case of a bridge collapse, more cars drive off into the water. As drivers multitask with, among other things, texting and phone calls, the risks become greater.

At peak times, getting off the road and onto other roads is mostly a random decision because one road can take just about the same time as another. Perhaps we should look at the probability of maintaining a certain speed on a road. We know that a density of 100 percent implies bumper-to-bumper traffic, which doesn’t move. At a density much above 28.6 percent, traffic slows and speed should be highly correlated to the density and the time of day. I should think that calculating probabilities would be easier now that we have GPS apps, Google Earth, and many satellite images to draw on.

Ronald S. Csuha
New York, NY