An O’Reilly blog post on smart cities praised technology that helps drivers find a ways to route around a traffic jam, reducing pollution. But those algorithmically-discovered back routes, formerly known only only to locals also route impatient drivers through neighborhood streets that were carefully traffic-calmed. The software, like much of the road system, was designed with the goal of efficiently moving cars, blind to the side effects.
One answer is more data – program speed limits into the software, and as cars become more automated, eventually slow down the car.
Another answer is more data and different assumptions. The software takes driving in traffic as a given. The software should know that the roads are jammed, and should be able to predict that the roads are going to be jammed at that hour. And then it could recommend not only an alternate route, but an alternative mode.