Snarled traffic is the result of too many people trying to go to the same place at the same time via the same route. In an attempt to save a few precious minutes, drivers will almost always choose what they consider to be the fastest route to their destination, ignoring other options that may be only a few minutes slower. (Because, why would you ever intentionally choose a slower route, right? You're very busy and important!)
Problem is, now that everybody and their dog has a GPS system in their car or phone (with today’s versions offering constant real time traffic updates), lots and lots of drivers attempt the same route optimization. The frequent result: suboptimal system configuration. That’s a real science-y way of saying that you end up with an increase in congestion that's worse than the likely state of things if everyone had just remained clueless.
Researchers at MIT have shown (in simulations, anyway) that if we were all willing to take a wider variety of coordinated routes that may not be optimized on an individual level, it would yield an overall reduction in congestion. Your route might not be the most direct, but you'd be helping most other drivers—and though it’s counterintuitive, probably yourself too—save a few minutes, because of a reduction in overall congestion (the additional travel time due to traffic compared with free flow conditions). In simulations of traffic conditions in cities such as San Francisco and Boston, congestion was reduced by as much as 30 percent. And all you have to do is be just a tiny bit considerate.
Marta González, Serdar Colak, and Antonio Lima at MIT used billions of anonymized cell phone location records to track commuters during peak morning rush hour in five cities, including Boston, the San Francisco Bay area, Rio de Janeiro, Lisbon, and Porto. Each simulated commuter is doing what real commuters do: trying to get from home to work as quickly and efficiently as possible. They’re driving in a state that the researchers refer to as “user equilibrium.” (But I’ll use my own technical term: selfish.)
Because simulations suggest that as much as 30 percent of the total time lost to congestion is caused solely by selfish routing, the more intelligent way to go would be what’s called “socially optimal” routing. Each driver would follow a path calculated to get him or her to their destination while minimizing congestion for all drivers simultaneously. There's a continuum between selfish routing and socially optimized routing; the MIT team rates drivers’ motives on a scale of 0 to 1, where zero corresponds to those who choose routes purely selfishly, and 1 is for drivers willing to put themselves at the mercy of a congestion-minimizing routing algorithm, taking a longer route for the good of all. The motive-measuring metric is the amount of “extra” time a driver is willing to spend following what he perceives to be a an indirect route in order to diminish congestion for everybody.
It turns out that relatively small changes in overall travel times for some drivers can lead to significant benefits for all drivers. And you don't even have to be totally selfless to make it work. Let's take a look at an example:
The figure on the right shows three possible routes from downtown San Francisco to the San Francisco airport. The purple route straight down the bay is the fastest, most “selfish.” It rates a zero. Ideally, it would take the driver 20 minutes. The socially optimal route on the far left, which rates a 1, would take the driver 25 minutes to traverse. Taking the little optional offshoot in the middle (with a value of 0.1) would add 2 minutes to the trip, for a total of 22 minutes. However, if a significant number of people followed that 0.1-rated route, the average trip length for everyone attempting to get to the airport from downtown would be reduced by 2 minutes. The upshot: a small fraction of drivers might spend an extra 5 to 7 minutes on the road, but most drivers would save a few minutes, and some drivers would save more than 10 minutes. “Some,” in the case of a city like San Francisco, means tens of thousands.
Still, the researchers point out that, “in the best-case scenario, time savings would be imperceptible for the majority of the drivers. From this, it is clear that such routing solutions cannot fix the traffic problem for individual drivers but rather would contribute to the city as a whole.” A few minutes saved every day by hundreds of thousands of vehicles would yield multiple benefits including significant reductions in fuel consumption, pollution, and wear on roads. And if you're among that small fraction of drivers who end up saving 10 minutes or more during your commute, it'll work out pretty well for you personally, too.
The researchers don't offer any specifics on the "implied routing application" that would actually do all of this. You could easily imagine it as an app, but what's more interesting is to consider how a system like this might be implemented throughout a fleet of autonomous cars. If you're not doing the driving yourself, a few minutes would probably make much less of a difference, and a connected autonomous car could automatically choose routes that are socially optimized. It may even be incentivized on a per-user basis, like a small reduction in road tolls for drivers who had to spend some extra time contributing to socially optimized routing. Will any of this ever happen? If so, it's not going to be soon, and it'll depend on the willingness of people to occasionally mildly inconvenience themselves for the benefit of strangers.
Yeah, it's doomed.
Evan Ackerman is a senior editor at IEEE Spectrum. Since 2007, he has written over 6,000 articles on robotics and technology. He has a degree in Martian geology and is excellent at playing bagpipes.