Build a road, make traffic worse

Every morning some cars drive from S to E. There are two ways round: through town A or through town B. One half of each route is a wide highway that always takes 45 minutes. The other half is a narrow road that gets slower the more cars use it (cars ÷ 100 minutes).

Now the city opens a brand-new road from A to B that takes no time at all. What could go wrong?

What is going on

Each driver only asks “which route is fastest for me, given what everyone else is doing?” With the shortcut open, the zigzag is always at least as fast for any one driver, so everyone piles onto both narrow roads. The traffic settles where no one can do better by switching alone, but that point is worse for everyone than the world without the shortcut. This is Braess’s paradox: individually sensible choices add up to a collectively worse outcome.

Try around 4,000 cars: the shortcut costs every driver 15 extra minutes. Below 3,000 cars it genuinely helps, and at 9,000 it stops mattering. It is not just a textbook trick: Seoul removing the Cheonggyecheon expressway and New York closing parts of Broadway to cars are often cited as real-life cases where losing a road did not make traffic worse.

I studied this at London Business School with my group, including when closing a road is the right call. The full analysis is in our report and slides.