Does closing roads cut delays?

Yes, because people do the wrong analysis.

The authors give a simple example of how this could play out: Imagine two routes to a destination, a short but narrow bridge and a longer but wider highway. Let’s also imagine that the combined travel times of all the drivers is shortest if half take the bridge and half take the highway. But because each driver is selfishly trying to seek the shortest route for himself, this doesn’t happen. At first, everyone will go for the bridge because it’s shorter. But then, as the bridge becomes backed up, more drivers start taking the highway, until the congestion on the bridge starts to clear up. At that point more drivers go back to the bridge, which then becomes backed up again. Eventually, the traffic flow settles into what’s called the Nash equilibrium (named for the beautifully minded mathematician), in which each route takes the same amount of time. But in this equilibrium the travel time is actually longer than the average time it would take if half of the drivers took each route.

Note that this still happens even if – indeed, especially if – all the drivers have perfect information about what all the other drivers are doing, such as with a GPS that gives real-time traffic updates.

The problem here is similar to the one of feedback, as anyone who has designed and tested gain/feedback circuitry can attest. There is an oscillation to the signal — an ebb and flow of traffic density. There is a delay in the time between the signal and the feedback, and at some point the delay is 180º out of phase, so you add to the problem rather than subtracting from it.

Note that the last quoted sentence is actually incorrect — the real-time traffic update information tells you where traffic is, not where it is going. If you knew that a lot of drivers were heading to the bridge and would be there in 15 minutes — about when you would arrive — you wouldn’t take the bridge. But all you know is how many are on the bridge right now. The information you are missing is how many drivers have made the decision to use the bridge.