The reason more roads can mean slower traffic

A city adds a road to relieve congestion. Journey times get worse. No engineer miscalculated, no driver behaved irrationally, and no unusual event occurred. The network simply produced an outcome that none of its users wanted and all of them helped create.

What everyone sees

Capacity was added, so throughput should rise. When it does not, the explanation reached for is volume: more people must be driving, or the work was done badly. Both are testable and both are frequently false.

What is actually happening

When every driver independently picks their own fastest route, the resulting distribution of traffic need not be the fastest for the network. Dietrich Braess showed in 1968 that adding a link can move the equilibrium to a worse one for everybody. The new road is individually attractive and collectively destructive, and the gap between those two facts is stable — it does not resolve with learning, because each driver is already choosing optimally. Seoul and New York have both recorded improved flow after closing roads.

Why it stays hidden

Because the failure has no author. Every participant made the right decision, so there is nobody to blame and no obvious error to point at. Systems that fail through correct individual behaviour are the hardest kind to see, and the intuition that capacity equals throughput is very hard to dislodge.